pi-free 2.2.4 → 2.2.6

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (35) hide show
  1. package/CHANGELOG.md +24 -0
  2. package/README.md +1 -1
  3. package/banner.svg +1 -1
  4. package/config.ts +58 -29
  5. package/constants.ts +1 -4
  6. package/index.ts +30 -24
  7. package/lib/built-in-toggle.ts +185 -18
  8. package/lib/provider-cache.ts +22 -0
  9. package/package.json +74 -74
  10. package/provider-failover/benchmark-lookup.ts +4 -1
  11. package/provider-failover/benchmarks-chunk-0.ts +570 -570
  12. package/provider-failover/benchmarks-chunk-1.ts +676 -676
  13. package/provider-failover/benchmarks-chunk-2.ts +673 -673
  14. package/provider-failover/benchmarks-chunk-3.ts +680 -680
  15. package/provider-failover/benchmarks-chunk-4.ts +683 -683
  16. package/provider-failover/benchmarks-chunk-5.ts +816 -474
  17. package/provider-helper.ts +64 -0
  18. package/providers/bai/bai.ts +8 -2
  19. package/providers/crofai/crofai.ts +8 -2
  20. package/providers/deepinfra/deepinfra.ts +8 -2
  21. package/providers/dynamic-built-in/index.ts +36 -26
  22. package/providers/kilo/kilo.ts +41 -22
  23. package/providers/novita/novita.ts +8 -2
  24. package/providers/openmodel/openmodel.ts +8 -2
  25. package/providers/qoder/models.ts +63 -175
  26. package/providers/qoder/qoder.ts +49 -84
  27. package/providers/qoder/stream.ts +182 -274
  28. package/providers/qoder/transform.ts +5 -2
  29. package/providers/routeway/routeway.ts +8 -2
  30. package/providers/sambanova/sambanova.ts +12 -6
  31. package/providers/together/together.ts +8 -2
  32. package/providers/tokenrouter/tokenrouter.ts +8 -2
  33. package/providers/zenmux/zenmux.ts +8 -2
  34. package/providers/codestral/codestral.ts +0 -128
  35. package/providers/qoder/encoding.ts +0 -48
package/provider-failover/benchmarks-chunk-1.ts CHANGED
@@ -1,20 +1,20 @@
1
1
  // Auto-generated benchmark data chunk 1
2
- // Models: lfm2-2.6b .. minicpm-v-4.6-1.3b (90 entries)
3
- // Last updated: 2026-06-01
2
+ // Models: phi-4 .. k2-v2-high (90 entries)
3
+ // Last updated: 2026-07-01
4
4
  // DO NOT EDIT MANUALLY — generated by scripts/update-benchmarks.ts
5
5
 
6
6
  import type { HardcodedBenchmark } from "./hardcoded-benchmarks.ts";
7
7
 
8
8
  export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
9
- "lfm2-2.6b": {
9
+ "phi-4": {
10
10
  // AA specific benchmarks
11
- codingIndex: 1.4,
12
- mathIndex: 8.3,
11
+ codingIndex: undefined,
12
+ mathIndex: 18,
13
13
 
14
14
  // Academic benchmarks
15
- mmluPro: 0.298,
16
- gpqa: 0.306,
17
- hle: 0.052,
15
+ mmluPro: 0.714,
16
+ gpqa: 0.575,
17
+ hle: 0.041,
18
18
 
19
19
  // Capabilities
20
20
  contextWindow: 8192,
@@ -22,18 +22,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
22
22
  supportsVision: false,
23
23
 
24
24
  // Metadata
25
- lastUpdated: "2026-06-01",
26
- originalModel: "LFM2 2.6B",
25
+ lastUpdated: "2026-07-01",
26
+ originalModel: "Phi-4",
27
27
  },
28
- "lfm2.5-1.2b-thinking": {
28
+ "phi-4-mini-instruct": {
29
29
  // AA specific benchmarks
30
- codingIndex: 1.4,
31
- mathIndex: undefined,
30
+ codingIndex: undefined,
31
+ mathIndex: 6.7,
32
32
 
33
33
  // Academic benchmarks
34
- mmluPro: undefined,
35
- gpqa: 0.339,
36
- hle: 0.061,
34
+ mmluPro: 0.465,
35
+ gpqa: 0.331,
36
+ hle: 0.042,
37
37
 
38
38
  // Capabilities
39
39
  contextWindow: 8192,
@@ -41,18 +41,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
41
41
  supportsVision: false,
42
42
 
43
43
  // Metadata
44
- lastUpdated: "2026-06-01",
45
- originalModel: "LFM2.5-1.2B-Thinking",
44
+ lastUpdated: "2026-07-01",
45
+ originalModel: "Phi-4 Mini Instruct",
46
46
  },
47
- "lfm2.5-1.2b-instruct": {
47
+ "phi-4-multimodal-instruct": {
48
48
  // AA specific benchmarks
49
- codingIndex: 0.8,
49
+ codingIndex: undefined,
50
50
  mathIndex: undefined,
51
51
 
52
52
  // Academic benchmarks
53
- mmluPro: undefined,
54
- gpqa: 0.326,
55
- hle: 0.068,
53
+ mmluPro: 0.485,
54
+ gpqa: 0.315,
55
+ hle: 0.044,
56
56
 
57
57
  // Capabilities
58
58
  contextWindow: 8192,
@@ -60,18 +60,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
60
60
  supportsVision: false,
61
61
 
62
62
  // Metadata
63
- lastUpdated: "2026-06-01",
64
- originalModel: "LFM2.5-1.2B-Instruct",
63
+ lastUpdated: "2026-07-01",
64
+ originalModel: "Phi-4 Multimodal Instruct",
65
65
  },
66
- "solar-pro-3": {
66
+ "lfm2.5-1.2b-thinking": {
67
67
  // AA specific benchmarks
68
- codingIndex: 13.3,
68
+ codingIndex: undefined,
69
69
  mathIndex: undefined,
70
70
 
71
71
  // Academic benchmarks
72
72
  mmluPro: undefined,
73
- gpqa: 0.724,
74
- hle: 0.101,
73
+ gpqa: 0.339,
74
+ hle: 0.061,
75
75
 
76
76
  // Capabilities
77
77
  contextWindow: 8192,
@@ -79,18 +79,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
79
79
  supportsVision: false,
80
80
 
81
81
  // Metadata
82
- lastUpdated: "2026-06-01",
83
- originalModel: "Solar Pro 3",
82
+ lastUpdated: "2026-07-01",
83
+ originalModel: "LFM2.5-1.2B-Thinking",
84
84
  },
85
- "solar-open-100b-reasoning": {
85
+ "lfm2.5-1.2b-instruct": {
86
86
  // AA specific benchmarks
87
- codingIndex: 10.5,
87
+ codingIndex: undefined,
88
88
  mathIndex: undefined,
89
89
 
90
90
  // Academic benchmarks
91
91
  mmluPro: undefined,
92
- gpqa: 0.657,
93
- hle: 0.092,
92
+ gpqa: 0.326,
93
+ hle: 0.068,
94
94
 
95
95
  // Capabilities
96
96
  contextWindow: 8192,
@@ -98,18 +98,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
98
98
  supportsVision: false,
99
99
 
100
100
  // Metadata
101
- lastUpdated: "2026-06-01",
102
- originalModel: "Solar Open 100B (Reasoning)",
101
+ lastUpdated: "2026-07-01",
102
+ originalModel: "LFM2.5-1.2B-Instruct",
103
103
  },
104
- "solar-pro-2-non-reasoning": {
104
+ "lfm2-8b-a1b": {
105
105
  // AA specific benchmarks
106
- codingIndex: 11.3,
107
- mathIndex: 30,
106
+ codingIndex: undefined,
107
+ mathIndex: 25.3,
108
108
 
109
109
  // Academic benchmarks
110
- mmluPro: 0.75,
111
- gpqa: 0.561,
112
- hle: 0.038,
110
+ mmluPro: 0.505,
111
+ gpqa: 0.344,
112
+ hle: 0.049,
113
113
 
114
114
  // Capabilities
115
115
  contextWindow: 8192,
@@ -117,18 +117,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
117
117
  supportsVision: false,
118
118
 
119
119
  // Metadata
120
- lastUpdated: "2026-06-01",
121
- originalModel: "Solar Pro 2 (Non-reasoning)",
120
+ lastUpdated: "2026-07-01",
121
+ originalModel: "LFM2 8B A1B",
122
122
  },
123
- "solar-pro-2-reasoning": {
123
+ "lfm2-2.6b": {
124
124
  // AA specific benchmarks
125
- codingIndex: 12.1,
126
- mathIndex: 61.3,
125
+ codingIndex: undefined,
126
+ mathIndex: 8.3,
127
127
 
128
128
  // Academic benchmarks
129
- mmluPro: 0.805,
130
- gpqa: 0.687,
131
- hle: 0.07,
129
+ mmluPro: 0.298,
130
+ gpqa: 0.306,
131
+ hle: 0.052,
132
132
 
133
133
  // Capabilities
134
134
  contextWindow: 8192,
@@ -136,18 +136,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
136
136
  supportsVision: false,
137
137
 
138
138
  // Metadata
139
- lastUpdated: "2026-06-01",
140
- originalModel: "Solar Pro 2 (Reasoning)",
139
+ lastUpdated: "2026-07-01",
140
+ originalModel: "LFM2 2.6B",
141
141
  },
142
- "minimax-m2.7": {
142
+ "lfm2.5-8b-a1b": {
143
143
  // AA specific benchmarks
144
- codingIndex: 41.9,
144
+ codingIndex: undefined,
145
145
  mathIndex: undefined,
146
146
 
147
147
  // Academic benchmarks
148
148
  mmluPro: undefined,
149
- gpqa: 0.874,
150
- hle: 0.281,
149
+ gpqa: 0.513,
150
+ hle: 0.069,
151
151
 
152
152
  // Capabilities
153
153
  contextWindow: 8192,
@@ -155,18 +155,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
155
155
  supportsVision: false,
156
156
 
157
157
  // Metadata
158
- lastUpdated: "2026-06-01",
159
- originalModel: "MiniMax-M2.7",
158
+ lastUpdated: "2026-07-01",
159
+ originalModel: "LFM2.5-8B-A1B",
160
160
  },
161
- "llama-3.1-nemotron-instruct-70b": {
161
+ "lfm2-24b-a2b": {
162
162
  // AA specific benchmarks
163
- codingIndex: 10.8,
164
- mathIndex: 11,
163
+ codingIndex: undefined,
164
+ mathIndex: undefined,
165
165
 
166
166
  // Academic benchmarks
167
- mmluPro: 0.69,
168
- gpqa: 0.465,
169
- hle: 0.046,
167
+ mmluPro: undefined,
168
+ gpqa: 0.474,
169
+ hle: 0.044,
170
170
 
171
171
  // Capabilities
172
172
  contextWindow: 8192,
@@ -174,18 +174,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
174
174
  supportsVision: false,
175
175
 
176
176
  // Metadata
177
- lastUpdated: "2026-06-01",
178
- originalModel: "Llama 3.1 Nemotron Instruct 70B",
177
+ lastUpdated: "2026-07-01",
178
+ originalModel: "LFM2 24B A2B",
179
179
  },
180
- "nvidia-nemotron-nano-12b-v2-vl-non-reasoning": {
180
+ "lfm2.5-vl-1.6b": {
181
181
  // AA specific benchmarks
182
- codingIndex: 5.9,
183
- mathIndex: 26.7,
182
+ codingIndex: undefined,
183
+ mathIndex: undefined,
184
184
 
185
185
  // Academic benchmarks
186
- mmluPro: 0.649,
187
- gpqa: 0.439,
188
- hle: 0.045,
186
+ mmluPro: undefined,
187
+ gpqa: 0.289,
188
+ hle: 0.051,
189
189
 
190
190
  // Capabilities
191
191
  contextWindow: 8192,
@@ -193,18 +193,37 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
193
193
  supportsVision: false,
194
194
 
195
195
  // Metadata
196
- lastUpdated: "2026-06-01",
197
- originalModel: "NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)",
196
+ lastUpdated: "2026-07-01",
197
+ originalModel: "LFM2.5-VL-1.6B",
198
198
  },
199
- "nvidia-nemotron-3-super-120b-a12b-reasoning": {
199
+ "solar-pro-2-reasoning": {
200
+ // AA specific benchmarks
201
+ codingIndex: undefined,
202
+ mathIndex: 61.3,
203
+
204
+ // Academic benchmarks
205
+ mmluPro: 0.805,
206
+ gpqa: 0.687,
207
+ hle: 0.07,
208
+
209
+ // Capabilities
210
+ contextWindow: 8192,
211
+ supportsReasoning: false,
212
+ supportsVision: false,
213
+
214
+ // Metadata
215
+ lastUpdated: "2026-07-01",
216
+ originalModel: "Solar Pro 2 (Reasoning)",
217
+ },
218
+ "solar-pro-3": {
200
219
  // AA specific benchmarks
201
- codingIndex: 31.2,
220
+ codingIndex: 16.2,
202
221
  mathIndex: undefined,
203
222
 
204
223
  // Academic benchmarks
205
224
  mmluPro: undefined,
206
- gpqa: 0.8,
207
- hle: 0.192,
225
+ gpqa: 0.724,
226
+ hle: 0.101,
208
227
 
209
228
  // Capabilities
210
229
  contextWindow: 8192,
@@ -212,18 +231,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
212
231
  supportsVision: false,
213
232
 
214
233
  // Metadata
215
- lastUpdated: "2026-06-01",
216
- originalModel: "NVIDIA Nemotron 3 Super 120B A12B (Reasoning)",
234
+ lastUpdated: "2026-07-01",
235
+ originalModel: "Solar Pro 3",
217
236
  },
218
- "nvidia-nemotron-3-nano-30b-a3b-non-reasoning": {
237
+ "solar-open-100b-reasoning": {
219
238
  // AA specific benchmarks
220
- codingIndex: 15.8,
221
- mathIndex: 13.3,
239
+ codingIndex: undefined,
240
+ mathIndex: undefined,
222
241
 
223
242
  // Academic benchmarks
224
- mmluPro: 0.579,
225
- gpqa: 0.399,
226
- hle: 0.046,
243
+ mmluPro: undefined,
244
+ gpqa: 0.657,
245
+ hle: 0.092,
227
246
 
228
247
  // Capabilities
229
248
  contextWindow: 8192,
@@ -231,18 +250,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
231
250
  supportsVision: false,
232
251
 
233
252
  // Metadata
234
- lastUpdated: "2026-06-01",
235
- originalModel: "NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)",
253
+ lastUpdated: "2026-07-01",
254
+ originalModel: "Solar Open 100B (Reasoning)",
236
255
  },
237
- "nvidia-nemotron-nano-9b-v2-non-reasoning": {
256
+ "solar-pro-2-non-reasoning": {
238
257
  // AA specific benchmarks
239
- codingIndex: 7.5,
240
- mathIndex: 62.3,
258
+ codingIndex: undefined,
259
+ mathIndex: 30,
241
260
 
242
261
  // Academic benchmarks
243
- mmluPro: 0.739,
244
- gpqa: 0.557,
245
- hle: 0.04,
262
+ mmluPro: 0.75,
263
+ gpqa: 0.561,
264
+ hle: 0.038,
246
265
 
247
266
  // Capabilities
248
267
  contextWindow: 8192,
@@ -250,18 +269,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
250
269
  supportsVision: false,
251
270
 
252
271
  // Metadata
253
- lastUpdated: "2026-06-01",
254
- originalModel: "NVIDIA Nemotron Nano 9B V2 (Non-reasoning)",
272
+ lastUpdated: "2026-07-01",
273
+ originalModel: "Solar Pro 2 (Non-reasoning)",
255
274
  },
256
- "nemotron-3-nano-omni-30b-a3b-reasoning": {
275
+ "minimax-m2.7": {
257
276
  // AA specific benchmarks
258
- codingIndex: 14.8,
277
+ codingIndex: 52.6,
259
278
  mathIndex: undefined,
260
279
 
261
280
  // Academic benchmarks
262
281
  mmluPro: undefined,
263
- gpqa: 0.469,
264
- hle: 0.053,
282
+ gpqa: 0.874,
283
+ hle: 0.281,
265
284
 
266
285
  // Capabilities
267
286
  contextWindow: 8192,
@@ -269,18 +288,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
269
288
  supportsVision: false,
270
289
 
271
290
  // Metadata
272
- lastUpdated: "2026-06-01",
273
- originalModel: "Nemotron 3 Nano Omni 30B A3B Reasoning",
291
+ lastUpdated: "2026-07-01",
292
+ originalModel: "MiniMax-M2.7",
274
293
  },
275
- "llama-nemotron-super-49b-v1.5-non-reasoning": {
294
+ "minimax-m3": {
276
295
  // AA specific benchmarks
277
- codingIndex: 10.5,
278
- mathIndex: 8,
296
+ codingIndex: 58.6,
297
+ mathIndex: undefined,
279
298
 
280
299
  // Academic benchmarks
281
- mmluPro: 0.692,
282
- gpqa: 0.481,
283
- hle: 0.043,
300
+ mmluPro: undefined,
301
+ gpqa: 0.929,
302
+ hle: 0.371,
284
303
 
285
304
  // Capabilities
286
305
  contextWindow: 8192,
@@ -288,18 +307,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
288
307
  supportsVision: false,
289
308
 
290
309
  // Metadata
291
- lastUpdated: "2026-06-01",
292
- originalModel: "Llama Nemotron Super 49B v1.5 (Non-reasoning)",
310
+ lastUpdated: "2026-07-01",
311
+ originalModel: "MiniMax-M3",
293
312
  },
294
- "nemotron-cascade-2-30b-a3b": {
313
+ "llama-3.1-nemotron-instruct-70b": {
295
314
  // AA specific benchmarks
296
- codingIndex: 25.8,
297
- mathIndex: undefined,
315
+ codingIndex: undefined,
316
+ mathIndex: 11,
298
317
 
299
318
  // Academic benchmarks
300
- mmluPro: undefined,
301
- gpqa: 0.758,
302
- hle: 0.114,
319
+ mmluPro: 0.69,
320
+ gpqa: 0.465,
321
+ hle: 0.046,
303
322
 
304
323
  // Capabilities
305
324
  contextWindow: 8192,
@@ -307,18 +326,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
307
326
  supportsVision: false,
308
327
 
309
328
  // Metadata
310
- lastUpdated: "2026-06-01",
311
- originalModel: "Nemotron Cascade 2 30B A3B",
329
+ lastUpdated: "2026-07-01",
330
+ originalModel: "Llama 3.1 Nemotron Instruct 70B",
312
331
  },
313
- "nvidia-nemotron-nano-12b-v2-vl-reasoning": {
332
+ "nvidia-nemotron-3-super-120b-a12b-reasoning": {
314
333
  // AA specific benchmarks
315
- codingIndex: 11.7,
316
- mathIndex: 75,
334
+ codingIndex: 37.7,
335
+ mathIndex: undefined,
317
336
 
318
337
  // Academic benchmarks
319
- mmluPro: 0.759,
320
- gpqa: 0.572,
321
- hle: 0.053,
338
+ mmluPro: undefined,
339
+ gpqa: 0.8,
340
+ hle: 0.192,
322
341
 
323
342
  // Capabilities
324
343
  contextWindow: 8192,
@@ -326,18 +345,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
326
345
  supportsVision: false,
327
346
 
328
347
  // Metadata
329
- lastUpdated: "2026-06-01",
330
- originalModel: "NVIDIA Nemotron Nano 12B v2 VL (Reasoning)",
348
+ lastUpdated: "2026-07-01",
349
+ originalModel: "NVIDIA Nemotron 3 Super 120B A12B (Reasoning)",
331
350
  },
332
- "nvidia-nemotron-3-nano-4b": {
351
+ "llama-65b": {
333
352
  // AA specific benchmarks
334
- codingIndex: 10,
353
+ codingIndex: undefined,
335
354
  mathIndex: undefined,
336
355
 
337
356
  // Academic benchmarks
338
357
  mmluPro: undefined,
339
- gpqa: 0.513,
340
- hle: 0.048,
358
+ gpqa: undefined,
359
+ hle: undefined,
341
360
 
342
361
  // Capabilities
343
362
  contextWindow: 8192,
@@ -345,18 +364,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
345
364
  supportsVision: false,
346
365
 
347
366
  // Metadata
348
- lastUpdated: "2026-06-01",
349
- originalModel: "NVIDIA Nemotron 3 Nano 4B",
367
+ lastUpdated: "2026-07-01",
368
+ originalModel: "Llama 65B",
350
369
  },
351
- "nvidia-nemotron-3-nano-30b-a3b-reasoning": {
370
+ "nemotron-3-ultra-550b-a55b-reasoning": {
352
371
  // AA specific benchmarks
353
- codingIndex: 19,
354
- mathIndex: 91,
372
+ codingIndex: 49.3,
373
+ mathIndex: undefined,
355
374
 
356
375
  // Academic benchmarks
357
- mmluPro: 0.794,
358
- gpqa: 0.757,
359
- hle: 0.102,
376
+ mmluPro: undefined,
377
+ gpqa: 0.867,
378
+ hle: 0.266,
360
379
 
361
380
  // Capabilities
362
381
  contextWindow: 8192,
@@ -364,18 +383,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
364
383
  supportsVision: false,
365
384
 
366
385
  // Metadata
367
- lastUpdated: "2026-06-01",
368
- originalModel: "NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)",
386
+ lastUpdated: "2026-07-01",
387
+ originalModel: "Nemotron 3 Ultra 550B A55B (Reasoning)",
369
388
  },
370
- "nvidia-nemotron-nano-9b-v2-reasoning": {
389
+ "nvidia-nemotron-nano-12b-v2-vl-non-reasoning": {
371
390
  // AA specific benchmarks
372
- codingIndex: 8.3,
373
- mathIndex: 69.7,
391
+ codingIndex: undefined,
392
+ mathIndex: 26.7,
374
393
 
375
394
  // Academic benchmarks
376
- mmluPro: 0.742,
377
- gpqa: 0.57,
378
- hle: 0.046,
395
+ mmluPro: 0.649,
396
+ gpqa: 0.439,
397
+ hle: 0.045,
379
398
 
380
399
  // Capabilities
381
400
  contextWindow: 8192,
@@ -383,18 +402,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
383
402
  supportsVision: false,
384
403
 
385
404
  // Metadata
386
- lastUpdated: "2026-06-01",
387
- originalModel: "NVIDIA Nemotron Nano 9B V2 (Reasoning)",
405
+ lastUpdated: "2026-07-01",
406
+ originalModel: "NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)",
388
407
  },
389
- "llama-nemotron-super-49b-v1.5-reasoning": {
408
+ "llama-nemotron-super-49b-v1.5-non-reasoning": {
390
409
  // AA specific benchmarks
391
- codingIndex: 15.1,
392
- mathIndex: 76.7,
410
+ codingIndex: undefined,
411
+ mathIndex: 8,
393
412
 
394
413
  // Academic benchmarks
395
- mmluPro: 0.814,
396
- gpqa: 0.748,
397
- hle: 0.068,
414
+ mmluPro: 0.692,
415
+ gpqa: 0.481,
416
+ hle: 0.043,
398
417
 
399
418
  // Capabilities
400
419
  contextWindow: 8192,
@@ -402,12 +421,12 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
402
421
  supportsVision: false,
403
422
 
404
423
  // Metadata
405
- lastUpdated: "2026-06-01",
406
- originalModel: "Llama Nemotron Super 49B v1.5 (Reasoning)",
424
+ lastUpdated: "2026-07-01",
425
+ originalModel: "Llama Nemotron Super 49B v1.5 (Non-reasoning)",
407
426
  },
408
427
  "llama-3.1-nemotron-ultra-253b-v1-reasoning": {
409
428
  // AA specific benchmarks
410
- codingIndex: 13.1,
429
+ codingIndex: undefined,
411
430
  mathIndex: 63.7,
412
431
 
413
432
  // Academic benchmarks
@@ -421,18 +440,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
421
440
  supportsVision: false,
422
441
 
423
442
  // Metadata
424
- lastUpdated: "2026-06-01",
443
+ lastUpdated: "2026-07-01",
425
444
  originalModel: "Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)",
426
445
  },
427
- "kimi-k2.6-non-reasoning": {
446
+ "nemotron-3-nano-omni-30b-a3b-reasoning": {
428
447
  // AA specific benchmarks
429
- codingIndex: 38.4,
448
+ codingIndex: undefined,
430
449
  mathIndex: undefined,
431
450
 
432
451
  // Academic benchmarks
433
452
  mmluPro: undefined,
434
- gpqa: 0.788,
435
- hle: 0.182,
453
+ gpqa: 0.469,
454
+ hle: 0.053,
436
455
 
437
456
  // Capabilities
438
457
  contextWindow: 8192,
@@ -440,18 +459,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
440
459
  supportsVision: false,
441
460
 
442
461
  // Metadata
443
- lastUpdated: "2026-06-01",
444
- originalModel: "Kimi K2.6 (Non-reasoning)",
462
+ lastUpdated: "2026-07-01",
463
+ originalModel: "Nemotron 3 Nano Omni 30B A3B Reasoning",
445
464
  },
446
- "kimi-k2.6": {
465
+ "nvidia-nemotron-3-nano-30b-a3b-non-reasoning": {
447
466
  // AA specific benchmarks
448
- codingIndex: 47.1,
449
- mathIndex: undefined,
467
+ codingIndex: undefined,
468
+ mathIndex: 13.3,
450
469
 
451
470
  // Academic benchmarks
452
- mmluPro: undefined,
453
- gpqa: 0.911,
454
- hle: 0.359,
471
+ mmluPro: 0.579,
472
+ gpqa: 0.399,
473
+ hle: 0.046,
455
474
 
456
475
  // Capabilities
457
476
  contextWindow: 8192,
@@ -459,37 +478,37 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
459
478
  supportsVision: false,
460
479
 
461
480
  // Metadata
462
- lastUpdated: "2026-06-01",
463
- originalModel: "Kimi K2.6",
481
+ lastUpdated: "2026-07-01",
482
+ originalModel: "NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)",
464
483
  },
465
- "kimi-linear-48b-a3b-instruct": {
484
+ "nvidia-nemotron-nano-12b-v2-vl-reasoning": {
466
485
  // AA specific benchmarks
467
- codingIndex: 14.2,
468
- mathIndex: 36.3,
486
+ codingIndex: undefined,
487
+ mathIndex: 75,
469
488
 
470
489
  // Academic benchmarks
471
- mmluPro: 0.585,
472
- gpqa: 0.412,
473
- hle: 0.027,
474
-
490
+ mmluPro: 0.759,
491
+ gpqa: 0.572,
492
+ hle: 0.053,
493
+
475
494
  // Capabilities
476
495
  contextWindow: 8192,
477
496
  supportsReasoning: false,
478
497
  supportsVision: false,
479
498
 
480
499
  // Metadata
481
- lastUpdated: "2026-06-01",
482
- originalModel: "Kimi Linear 48B A3B Instruct",
500
+ lastUpdated: "2026-07-01",
501
+ originalModel: "NVIDIA Nemotron Nano 12B v2 VL (Reasoning)",
483
502
  },
484
- "llama-65b": {
503
+ "nvidia-nemotron-nano-9b-v2-reasoning": {
485
504
  // AA specific benchmarks
486
505
  codingIndex: undefined,
487
- mathIndex: undefined,
506
+ mathIndex: 69.7,
488
507
 
489
508
  // Academic benchmarks
490
- mmluPro: undefined,
491
- gpqa: undefined,
492
- hle: undefined,
509
+ mmluPro: 0.742,
510
+ gpqa: 0.57,
511
+ hle: 0.046,
493
512
 
494
513
  // Capabilities
495
514
  contextWindow: 8192,
@@ -497,18 +516,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
497
516
  supportsVision: false,
498
517
 
499
518
  // Metadata
500
- lastUpdated: "2026-06-01",
501
- originalModel: "Llama 65B",
519
+ lastUpdated: "2026-07-01",
520
+ originalModel: "NVIDIA Nemotron Nano 9B V2 (Reasoning)",
502
521
  },
503
- "step-3.5-flash-2603": {
522
+ "nemotron-cascade-2-30b-a3b": {
504
523
  // AA specific benchmarks
505
- codingIndex: 34.6,
524
+ codingIndex: undefined,
506
525
  mathIndex: undefined,
507
526
 
508
527
  // Academic benchmarks
509
528
  mmluPro: undefined,
510
- gpqa: 0.826,
511
- hle: 0.226,
529
+ gpqa: 0.758,
530
+ hle: 0.114,
512
531
 
513
532
  // Capabilities
514
533
  contextWindow: 8192,
@@ -516,18 +535,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
516
535
  supportsVision: false,
517
536
 
518
537
  // Metadata
519
- lastUpdated: "2026-06-01",
520
- originalModel: "Step 3.5 Flash 2603",
538
+ lastUpdated: "2026-07-01",
539
+ originalModel: "Nemotron Cascade 2 30B A3B",
521
540
  },
522
- "step3-vl-10b": {
541
+ "llama-nemotron-super-49b-v1.5-reasoning": {
523
542
  // AA specific benchmarks
524
- codingIndex: 13.9,
525
- mathIndex: undefined,
543
+ codingIndex: undefined,
544
+ mathIndex: 76.7,
526
545
 
527
546
  // Academic benchmarks
528
- mmluPro: undefined,
529
- gpqa: 0.69,
530
- hle: 0.102,
547
+ mmluPro: 0.814,
548
+ gpqa: 0.748,
549
+ hle: 0.068,
531
550
 
532
551
  // Capabilities
533
552
  contextWindow: 8192,
@@ -535,18 +554,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
535
554
  supportsVision: false,
536
555
 
537
556
  // Metadata
538
- lastUpdated: "2026-06-01",
539
- originalModel: "Step3 VL 10B",
557
+ lastUpdated: "2026-07-01",
558
+ originalModel: "Llama Nemotron Super 49B v1.5 (Reasoning)",
540
559
  },
541
- "molmo-7b-d": {
560
+ "nvidia-nemotron-nano-9b-v2-non-reasoning": {
542
561
  // AA specific benchmarks
543
- codingIndex: 1.2,
544
- mathIndex: 0,
562
+ codingIndex: undefined,
563
+ mathIndex: 62.3,
545
564
 
546
565
  // Academic benchmarks
547
- mmluPro: 0.371,
548
- gpqa: 0.24,
549
- hle: 0.051,
566
+ mmluPro: 0.739,
567
+ gpqa: 0.557,
568
+ hle: 0.04,
550
569
 
551
570
  // Capabilities
552
571
  contextWindow: 8192,
@@ -554,18 +573,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
554
573
  supportsVision: false,
555
574
 
556
575
  // Metadata
557
- lastUpdated: "2026-06-01",
558
- originalModel: "Molmo 7B-D",
576
+ lastUpdated: "2026-07-01",
577
+ originalModel: "NVIDIA Nemotron Nano 9B V2 (Non-reasoning)",
559
578
  },
560
- "molmo2-8b": {
579
+ "nvidia-nemotron-3-nano-4b": {
561
580
  // AA specific benchmarks
562
- codingIndex: 4.4,
581
+ codingIndex: undefined,
563
582
  mathIndex: undefined,
564
583
 
565
584
  // Academic benchmarks
566
585
  mmluPro: undefined,
567
- gpqa: 0.425,
568
- hle: 0.044,
586
+ gpqa: 0.513,
587
+ hle: 0.048,
569
588
 
570
589
  // Capabilities
571
590
  contextWindow: 8192,
@@ -573,18 +592,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
573
592
  supportsVision: false,
574
593
 
575
594
  // Metadata
576
- lastUpdated: "2026-06-01",
577
- originalModel: "Molmo2-8B",
595
+ lastUpdated: "2026-07-01",
596
+ originalModel: "NVIDIA Nemotron 3 Nano 4B",
578
597
  },
579
- "olmo-3-7b-instruct": {
598
+ "nvidia-nemotron-3-nano-30b-a3b-reasoning": {
580
599
  // AA specific benchmarks
581
- codingIndex: 3.4,
582
- mathIndex: 41.3,
600
+ codingIndex: 14.4,
601
+ mathIndex: 91,
583
602
 
584
603
  // Academic benchmarks
585
- mmluPro: 0.522,
586
- gpqa: 0.4,
587
- hle: 0.058,
604
+ mmluPro: 0.794,
605
+ gpqa: 0.757,
606
+ hle: 0.102,
588
607
 
589
608
  // Capabilities
590
609
  contextWindow: 8192,
@@ -592,18 +611,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
592
611
  supportsVision: false,
593
612
 
594
613
  // Metadata
595
- lastUpdated: "2026-06-01",
596
- originalModel: "Olmo 3 7B Instruct",
614
+ lastUpdated: "2026-07-01",
615
+ originalModel: "NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)",
597
616
  },
598
- "olmo-3-7b-think": {
617
+ "kimi-k2.6-non-reasoning": {
599
618
  // AA specific benchmarks
600
- codingIndex: 7.6,
601
- mathIndex: 70.7,
619
+ codingIndex: undefined,
620
+ mathIndex: undefined,
602
621
 
603
622
  // Academic benchmarks
604
- mmluPro: 0.655,
605
- gpqa: 0.516,
606
- hle: 0.057,
623
+ mmluPro: undefined,
624
+ gpqa: 0.788,
625
+ hle: 0.182,
607
626
 
608
627
  // Capabilities
609
628
  contextWindow: 8192,
@@ -611,18 +630,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
611
630
  supportsVision: false,
612
631
 
613
632
  // Metadata
614
- lastUpdated: "2026-06-01",
615
- originalModel: "Olmo 3 7B Think",
633
+ lastUpdated: "2026-07-01",
634
+ originalModel: "Kimi K2.6 (Non-reasoning)",
616
635
  },
617
- "olmo-3.1-32b-instruct": {
636
+ "kimi-k2.7-code": {
618
637
  // AA specific benchmarks
619
- codingIndex: 5.6,
638
+ codingIndex: 60.8,
620
639
  mathIndex: undefined,
621
640
 
622
641
  // Academic benchmarks
623
642
  mmluPro: undefined,
624
- gpqa: 0.539,
625
- hle: 0.049,
643
+ gpqa: 0.896,
644
+ hle: 0.328,
626
645
 
627
646
  // Capabilities
628
647
  contextWindow: 8192,
@@ -630,18 +649,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
630
649
  supportsVision: false,
631
650
 
632
651
  // Metadata
633
- lastUpdated: "2026-06-01",
634
- originalModel: "Olmo 3.1 32B Instruct",
652
+ lastUpdated: "2026-07-01",
653
+ originalModel: "Kimi K2.7 Code",
635
654
  },
636
- "olmo-3.1-32b-think": {
655
+ "kimi-k2.6": {
637
656
  // AA specific benchmarks
638
- codingIndex: 9.8,
639
- mathIndex: 77.3,
657
+ codingIndex: 56,
658
+ mathIndex: undefined,
640
659
 
641
660
  // Academic benchmarks
642
- mmluPro: 0.763,
643
- gpqa: 0.591,
644
- hle: 0.06,
661
+ mmluPro: undefined,
662
+ gpqa: 0.911,
663
+ hle: 0.359,
645
664
 
646
665
  // Capabilities
647
666
  contextWindow: 8192,
@@ -649,18 +668,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
649
668
  supportsVision: false,
650
669
 
651
670
  // Metadata
652
- lastUpdated: "2026-06-01",
653
- originalModel: "Olmo 3.1 32B Think",
671
+ lastUpdated: "2026-07-01",
672
+ originalModel: "Kimi K2.6",
654
673
  },
655
- "granite-4.1-8b": {
674
+ "qwen-chat-14b": {
656
675
  // AA specific benchmarks
657
- codingIndex: 7.3,
676
+ codingIndex: undefined,
658
677
  mathIndex: undefined,
659
678
 
660
679
  // Academic benchmarks
661
680
  mmluPro: undefined,
662
- gpqa: 0.433,
663
- hle: 0.038,
681
+ gpqa: undefined,
682
+ hle: undefined,
664
683
 
665
684
  // Capabilities
666
685
  contextWindow: 8192,
@@ -668,18 +687,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
668
687
  supportsVision: false,
669
688
 
670
689
  // Metadata
671
- lastUpdated: "2026-06-01",
672
- originalModel: "Granite 4.1 8B",
690
+ lastUpdated: "2026-07-01",
691
+ originalModel: "Qwen Chat 14B",
673
692
  },
674
- "granite-4.0-350m": {
693
+ "kimi-linear-48b-a3b-instruct": {
675
694
  // AA specific benchmarks
676
- codingIndex: 0.3,
677
- mathIndex: 0,
695
+ codingIndex: undefined,
696
+ mathIndex: 36.3,
678
697
 
679
698
  // Academic benchmarks
680
- mmluPro: 0.124,
681
- gpqa: 0.261,
682
- hle: 0.057,
699
+ mmluPro: 0.585,
700
+ gpqa: 0.412,
701
+ hle: 0.027,
683
702
 
684
703
  // Capabilities
685
704
  contextWindow: 8192,
@@ -687,37 +706,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
687
706
  supportsVision: false,
688
707
 
689
708
  // Metadata
690
- lastUpdated: "2026-06-01",
691
- originalModel: "Granite 4.0 350M",
709
+ lastUpdated: "2026-07-01",
710
+ originalModel: "Kimi Linear 48B A3B Instruct",
692
711
  },
693
- "granite-4.1-3b": {
712
+ "step-3.7-flash": {
694
713
  // AA specific benchmarks
695
- codingIndex: 5.5,
714
+ codingIndex: 37.3,
696
715
  mathIndex: undefined,
697
716
 
698
717
  // Academic benchmarks
699
718
  mmluPro: undefined,
700
- gpqa: 0.314,
701
- hle: 0.034,
702
-
703
- // Capabilities
704
- contextWindow: 8192,
705
- supportsReasoning: false,
706
- supportsVision: false,
707
-
708
- // Metadata
709
- lastUpdated: "2026-06-01",
710
- originalModel: "Granite 4.1 3B",
711
- },
712
- "granite-4.0-1b": {
713
- // AA specific benchmarks
714
- codingIndex: 2.9,
715
- mathIndex: 6.3,
716
-
717
- // Academic benchmarks
718
- mmluPro: 0.325,
719
- gpqa: 0.281,
720
- hle: 0.051,
719
+ gpqa: 0.809,
720
+ hle: 0.199,
721
721
 
722
722
  // Capabilities
723
723
  contextWindow: 8192,
@@ -725,18 +725,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
725
725
  supportsVision: false,
726
726
 
727
727
  // Metadata
728
- lastUpdated: "2026-06-01",
729
- originalModel: "Granite 4.0 1B",
728
+ lastUpdated: "2026-07-01",
729
+ originalModel: "Step 3.7 Flash",
730
730
  },
731
- "granite-4.0-h-350m": {
731
+ "step3-vl-10b": {
732
732
  // AA specific benchmarks
733
- codingIndex: 0.6,
734
- mathIndex: 1.3,
733
+ codingIndex: undefined,
734
+ mathIndex: undefined,
735
735
 
736
736
  // Academic benchmarks
737
- mmluPro: 0.127,
738
- gpqa: 0.257,
739
- hle: 0.064,
737
+ mmluPro: undefined,
738
+ gpqa: 0.69,
739
+ hle: 0.102,
740
740
 
741
741
  // Capabilities
742
742
  contextWindow: 8192,
@@ -744,18 +744,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
744
744
  supportsVision: false,
745
745
 
746
746
  // Metadata
747
- lastUpdated: "2026-06-01",
748
- originalModel: "Granite 4.0 H 350M",
747
+ lastUpdated: "2026-07-01",
748
+ originalModel: "Step3 VL 10B",
749
749
  },
750
- "granite-4.1-30b": {
750
+ "step-3.5-flash-2603": {
751
751
  // AA specific benchmarks
752
- codingIndex: 10.1,
752
+ codingIndex: undefined,
753
753
  mathIndex: undefined,
754
754
 
755
755
  // Academic benchmarks
756
756
  mmluPro: undefined,
757
- gpqa: 0.481,
758
- hle: 0.042,
757
+ gpqa: 0.826,
758
+ hle: 0.226,
759
759
 
760
760
  // Capabilities
761
761
  contextWindow: 8192,
@@ -763,18 +763,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
763
763
  supportsVision: false,
764
764
 
765
765
  // Metadata
766
- lastUpdated: "2026-06-01",
767
- originalModel: "Granite 4.1 30B",
766
+ lastUpdated: "2026-07-01",
767
+ originalModel: "Step 3.5 Flash 2603",
768
768
  },
769
- "granite-4.0-h-1b": {
769
+ "olmo-3.1-32b-think": {
770
770
  // AA specific benchmarks
771
- codingIndex: 2.7,
772
- mathIndex: 6.3,
771
+ codingIndex: undefined,
772
+ mathIndex: 77.3,
773
773
 
774
774
  // Academic benchmarks
775
- mmluPro: 0.277,
776
- gpqa: 0.263,
777
- hle: 0.05,
775
+ mmluPro: 0.763,
776
+ gpqa: 0.591,
777
+ hle: 0.06,
778
778
 
779
779
  // Capabilities
780
780
  contextWindow: 8192,
@@ -782,18 +782,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
782
782
  supportsVision: false,
783
783
 
784
784
  // Metadata
785
- lastUpdated: "2026-06-01",
786
- originalModel: "Granite 4.0 H 1B",
785
+ lastUpdated: "2026-07-01",
786
+ originalModel: "Olmo 3.1 32B Think",
787
787
  },
788
- "granite-4.0-h-small": {
788
+ "molmo2-8b": {
789
789
  // AA specific benchmarks
790
- codingIndex: 8.5,
791
- mathIndex: 13.7,
790
+ codingIndex: undefined,
791
+ mathIndex: undefined,
792
792
 
793
793
  // Academic benchmarks
794
- mmluPro: 0.624,
795
- gpqa: 0.416,
796
- hle: 0.037,
794
+ mmluPro: undefined,
795
+ gpqa: 0.425,
796
+ hle: 0.044,
797
797
 
798
798
  // Capabilities
799
799
  contextWindow: 8192,
@@ -801,18 +801,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
801
801
  supportsVision: false,
802
802
 
803
803
  // Metadata
804
- lastUpdated: "2026-06-01",
805
- originalModel: "Granite 4.0 H Small",
804
+ lastUpdated: "2026-07-01",
805
+ originalModel: "Molmo2-8B",
806
806
  },
807
- "granite-4.0-micro": {
807
+ "olmo-3-7b-instruct": {
808
808
  // AA specific benchmarks
809
- codingIndex: 5,
810
- mathIndex: 6,
809
+ codingIndex: undefined,
810
+ mathIndex: 41.3,
811
811
 
812
812
  // Academic benchmarks
813
- mmluPro: 0.447,
814
- gpqa: 0.336,
815
- hle: 0.051,
813
+ mmluPro: 0.522,
814
+ gpqa: 0.4,
815
+ hle: 0.058,
816
816
 
817
817
  // Capabilities
818
818
  contextWindow: 8192,
@@ -820,18 +820,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
820
820
  supportsVision: false,
821
821
 
822
822
  // Metadata
823
- lastUpdated: "2026-06-01",
824
- originalModel: "Granite 4.0 Micro",
823
+ lastUpdated: "2026-07-01",
824
+ originalModel: "Olmo 3 7B Instruct",
825
825
  },
826
- "mercury-2": {
826
+ "olmo-3.1-32b-instruct": {
827
827
  // AA specific benchmarks
828
- codingIndex: 30.6,
828
+ codingIndex: undefined,
829
829
  mathIndex: undefined,
830
830
 
831
831
  // Academic benchmarks
832
832
  mmluPro: undefined,
833
- gpqa: 0.77,
834
- hle: 0.155,
833
+ gpqa: 0.539,
834
+ hle: 0.049,
835
835
 
836
836
  // Capabilities
837
837
  contextWindow: 8192,
@@ -839,17 +839,17 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
839
839
  supportsVision: false,
840
840
 
841
841
  // Metadata
842
- lastUpdated: "2026-06-01",
843
- originalModel: "Mercury 2",
842
+ lastUpdated: "2026-07-01",
843
+ originalModel: "Olmo 3.1 32B Instruct",
844
844
  },
845
- "reka-flash-3": {
845
+ "molmo-7b-d": {
846
846
  // AA specific benchmarks
847
- codingIndex: 8.9,
848
- mathIndex: 33.7,
847
+ codingIndex: undefined,
848
+ mathIndex: 0,
849
849
 
850
850
  // Academic benchmarks
851
- mmluPro: 0.669,
852
- gpqa: 0.529,
851
+ mmluPro: 0.371,
852
+ gpqa: 0.24,
853
853
  hle: 0.051,
854
854
 
855
855
  // Capabilities
@@ -858,18 +858,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
858
858
  supportsVision: false,
859
859
 
860
860
  // Metadata
861
- lastUpdated: "2026-06-01",
862
- originalModel: "Reka Flash 3",
861
+ lastUpdated: "2026-07-01",
862
+ originalModel: "Molmo 7B-D",
863
863
  },
864
- "deephermes-3---llama-3.1-8b-preview-non-reasoning": {
864
+ "olmo-3-7b-think": {
865
865
  // AA specific benchmarks
866
866
  codingIndex: undefined,
867
- mathIndex: undefined,
867
+ mathIndex: 70.7,
868
868
 
869
869
  // Academic benchmarks
870
- mmluPro: 0.365,
871
- gpqa: 0.27,
872
- hle: 0.043,
870
+ mmluPro: 0.655,
871
+ gpqa: 0.516,
872
+ hle: 0.057,
873
873
 
874
874
  // Capabilities
875
875
  contextWindow: 8192,
@@ -877,18 +877,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
877
877
  supportsVision: false,
878
878
 
879
879
  // Metadata
880
- lastUpdated: "2026-06-01",
881
- originalModel: "DeepHermes 3 - Llama-3.1 8B Preview (Non-reasoning)",
880
+ lastUpdated: "2026-07-01",
881
+ originalModel: "Olmo 3 7B Think",
882
882
  },
883
- "hermes-4---llama-3.1-405b-reasoning": {
883
+ "granite-4.1-8b": {
884
884
  // AA specific benchmarks
885
- codingIndex: 16,
886
- mathIndex: 69.7,
885
+ codingIndex: 9.5,
886
+ mathIndex: undefined,
887
887
 
888
888
  // Academic benchmarks
889
- mmluPro: 0.829,
890
- gpqa: 0.727,
891
- hle: 0.103,
889
+ mmluPro: undefined,
890
+ gpqa: 0.433,
891
+ hle: 0.038,
892
892
 
893
893
  // Capabilities
894
894
  contextWindow: 8192,
@@ -896,18 +896,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
896
896
  supportsVision: false,
897
897
 
898
898
  // Metadata
899
- lastUpdated: "2026-06-01",
900
- originalModel: "Hermes 4 - Llama-3.1 405B (Reasoning)",
899
+ lastUpdated: "2026-07-01",
900
+ originalModel: "Granite 4.1 8B",
901
901
  },
902
- "hermes-4---llama-3.1-405b-non-reasoning": {
902
+ "granite-4.0-h-small": {
903
903
  // AA specific benchmarks
904
- codingIndex: 18.1,
905
- mathIndex: 15.3,
904
+ codingIndex: undefined,
905
+ mathIndex: 13.7,
906
906
 
907
907
  // Academic benchmarks
908
- mmluPro: 0.729,
909
- gpqa: 0.536,
910
- hle: 0.042,
908
+ mmluPro: 0.624,
909
+ gpqa: 0.416,
910
+ hle: 0.037,
911
911
 
912
912
  // Capabilities
913
913
  contextWindow: 8192,
@@ -915,18 +915,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
915
915
  supportsVision: false,
916
916
 
917
917
  // Metadata
918
- lastUpdated: "2026-06-01",
919
- originalModel: "Hermes 4 - Llama-3.1 405B (Non-reasoning)",
918
+ lastUpdated: "2026-07-01",
919
+ originalModel: "Granite 4.0 H Small",
920
920
  },
921
- "deephermes-3---mistral-24b-preview-non-reasoning": {
921
+ "granite-4.1-30b": {
922
922
  // AA specific benchmarks
923
- codingIndex: undefined,
923
+ codingIndex: 10.4,
924
924
  mathIndex: undefined,
925
925
 
926
926
  // Academic benchmarks
927
- mmluPro: 0.58,
928
- gpqa: 0.382,
929
- hle: 0.039,
927
+ mmluPro: undefined,
928
+ gpqa: 0.481,
929
+ hle: 0.042,
930
930
 
931
931
  // Capabilities
932
932
  contextWindow: 8192,
@@ -934,18 +934,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
934
934
  supportsVision: false,
935
935
 
936
936
  // Metadata
937
- lastUpdated: "2026-06-01",
938
- originalModel: "DeepHermes 3 - Mistral 24B Preview (Non-reasoning)",
937
+ lastUpdated: "2026-07-01",
938
+ originalModel: "Granite 4.1 30B",
939
939
  },
940
- "hermes-4---llama-3.1-70b-reasoning": {
940
+ "granite-4.0-h-350m": {
941
941
  // AA specific benchmarks
942
- codingIndex: 14.4,
943
- mathIndex: 68.7,
942
+ codingIndex: undefined,
943
+ mathIndex: 1.3,
944
944
 
945
945
  // Academic benchmarks
946
- mmluPro: 0.811,
947
- gpqa: 0.699,
948
- hle: 0.079,
946
+ mmluPro: 0.127,
947
+ gpqa: 0.257,
948
+ hle: 0.064,
949
949
 
950
950
  // Capabilities
951
951
  contextWindow: 8192,
@@ -953,18 +953,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
953
953
  supportsVision: false,
954
954
 
955
955
  // Metadata
956
- lastUpdated: "2026-06-01",
957
- originalModel: "Hermes 4 - Llama-3.1 70B (Reasoning)",
956
+ lastUpdated: "2026-07-01",
957
+ originalModel: "Granite 4.0 H 350M",
958
958
  },
959
- "hermes-4---llama-3.1-70b-non-reasoning": {
959
+ "granite-4.0-1b": {
960
960
  // AA specific benchmarks
961
- codingIndex: 9.2,
962
- mathIndex: 11.3,
961
+ codingIndex: undefined,
962
+ mathIndex: 6.3,
963
963
 
964
964
  // Academic benchmarks
965
- mmluPro: 0.664,
966
- gpqa: 0.491,
967
- hle: 0.036,
965
+ mmluPro: 0.325,
966
+ gpqa: 0.281,
967
+ hle: 0.051,
968
968
 
969
969
  // Capabilities
970
970
  contextWindow: 8192,
@@ -972,18 +972,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
972
972
  supportsVision: false,
973
973
 
974
974
  // Metadata
975
- lastUpdated: "2026-06-01",
976
- originalModel: "Hermes 4 - Llama-3.1 70B (Non-reasoning)",
975
+ lastUpdated: "2026-07-01",
976
+ originalModel: "Granite 4.0 1B",
977
977
  },
978
- "exaone-4.0-1.2b-reasoning": {
978
+ "granite-4.1-3b": {
979
979
  // AA specific benchmarks
980
- codingIndex: 3.1,
981
- mathIndex: 50.3,
980
+ codingIndex: undefined,
981
+ mathIndex: undefined,
982
982
 
983
983
  // Academic benchmarks
984
- mmluPro: 0.588,
985
- gpqa: 0.515,
986
- hle: 0.058,
984
+ mmluPro: undefined,
985
+ gpqa: 0.314,
986
+ hle: 0.034,
987
987
 
988
988
  // Capabilities
989
989
  contextWindow: 8192,
@@ -991,18 +991,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
991
991
  supportsVision: false,
992
992
 
993
993
  // Metadata
994
- lastUpdated: "2026-06-01",
995
- originalModel: "Exaone 4.0 1.2B (Reasoning)",
994
+ lastUpdated: "2026-07-01",
995
+ originalModel: "Granite 4.1 3B",
996
996
  },
997
- "exaone-4.0-32b-reasoning": {
997
+ "granite-4.0-350m": {
998
998
  // AA specific benchmarks
999
- codingIndex: 14,
1000
- mathIndex: 80,
999
+ codingIndex: undefined,
1000
+ mathIndex: 0,
1001
1001
 
1002
1002
  // Academic benchmarks
1003
- mmluPro: 0.818,
1004
- gpqa: 0.739,
1005
- hle: 0.105,
1003
+ mmluPro: 0.124,
1004
+ gpqa: 0.261,
1005
+ hle: 0.057,
1006
1006
 
1007
1007
  // Capabilities
1008
1008
  contextWindow: 8192,
@@ -1010,18 +1010,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1010
1010
  supportsVision: false,
1011
1011
 
1012
1012
  // Metadata
1013
- lastUpdated: "2026-06-01",
1014
- originalModel: "EXAONE 4.0 32B (Reasoning)",
1013
+ lastUpdated: "2026-07-01",
1014
+ originalModel: "Granite 4.0 350M",
1015
1015
  },
1016
- "k-exaone-reasoning": {
1016
+ "granite-4.0-micro": {
1017
1017
  // AA specific benchmarks
1018
- codingIndex: 27,
1019
- mathIndex: 90.3,
1018
+ codingIndex: undefined,
1019
+ mathIndex: 6,
1020
1020
 
1021
1021
  // Academic benchmarks
1022
- mmluPro: 0.838,
1023
- gpqa: 0.783,
1024
- hle: 0.131,
1022
+ mmluPro: 0.447,
1023
+ gpqa: 0.336,
1024
+ hle: 0.051,
1025
1025
 
1026
1026
  // Capabilities
1027
1027
  contextWindow: 8192,
@@ -1029,18 +1029,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1029
1029
  supportsVision: false,
1030
1030
 
1031
1031
  // Metadata
1032
- lastUpdated: "2026-06-01",
1033
- originalModel: "K-EXAONE (Reasoning)",
1032
+ lastUpdated: "2026-07-01",
1033
+ originalModel: "Granite 4.0 Micro",
1034
1034
  },
1035
- "k-exaone-non-reasoning": {
1035
+ "granite-4.0-h-1b": {
1036
1036
  // AA specific benchmarks
1037
- codingIndex: 13.5,
1038
- mathIndex: 44,
1037
+ codingIndex: undefined,
1038
+ mathIndex: 6.3,
1039
1039
 
1040
1040
  // Academic benchmarks
1041
- mmluPro: 0.81,
1042
- gpqa: 0.695,
1043
- hle: 0.054,
1041
+ mmluPro: 0.277,
1042
+ gpqa: 0.263,
1043
+ hle: 0.05,
1044
1044
 
1045
1045
  // Capabilities
1046
1046
  contextWindow: 8192,
@@ -1048,18 +1048,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1048
1048
  supportsVision: false,
1049
1049
 
1050
1050
  // Metadata
1051
- lastUpdated: "2026-06-01",
1052
- originalModel: "K-EXAONE (Non-reasoning)",
1051
+ lastUpdated: "2026-07-01",
1052
+ originalModel: "Granite 4.0 H 1B",
1053
1053
  },
1054
- "exaone-4.5-33b": {
1054
+ "mercury-2": {
1055
1055
  // AA specific benchmarks
1056
- codingIndex: 23,
1056
+ codingIndex: undefined,
1057
1057
  mathIndex: undefined,
1058
1058
 
1059
1059
  // Academic benchmarks
1060
1060
  mmluPro: undefined,
1061
- gpqa: 0.794,
1062
- hle: 0.116,
1061
+ gpqa: 0.77,
1062
+ hle: 0.155,
1063
1063
 
1064
1064
  // Capabilities
1065
1065
  contextWindow: 8192,
@@ -1067,18 +1067,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1067
1067
  supportsVision: false,
1068
1068
 
1069
1069
  // Metadata
1070
- lastUpdated: "2026-06-01",
1071
- originalModel: "EXAONE 4.5 33B",
1070
+ lastUpdated: "2026-07-01",
1071
+ originalModel: "Mercury 2",
1072
1072
  },
1073
- "exaone-4.0-32b-non-reasoning": {
1073
+ "reka-flash-3": {
1074
1074
  // AA specific benchmarks
1075
- codingIndex: 9.4,
1076
- mathIndex: 39.3,
1075
+ codingIndex: undefined,
1076
+ mathIndex: 33.7,
1077
1077
 
1078
1078
  // Academic benchmarks
1079
- mmluPro: 0.768,
1080
- gpqa: 0.628,
1081
- hle: 0.049,
1079
+ mmluPro: 0.669,
1080
+ gpqa: 0.529,
1081
+ hle: 0.051,
1082
1082
 
1083
1083
  // Capabilities
1084
1084
  contextWindow: 8192,
@@ -1086,18 +1086,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1086
1086
  supportsVision: false,
1087
1087
 
1088
1088
  // Metadata
1089
- lastUpdated: "2026-06-01",
1090
- originalModel: "EXAONE 4.0 32B (Non-reasoning)",
1089
+ lastUpdated: "2026-07-01",
1090
+ originalModel: "Reka Flash 3",
1091
1091
  },
1092
- "exaone-4.0-1.2b-non-reasoning": {
1092
+ "hermes-4---llama-3.1-70b-non-reasoning": {
1093
1093
  // AA specific benchmarks
1094
- codingIndex: 2.5,
1095
- mathIndex: 24,
1094
+ codingIndex: undefined,
1095
+ mathIndex: 11.3,
1096
1096
 
1097
1097
  // Academic benchmarks
1098
- mmluPro: 0.5,
1099
- gpqa: 0.424,
1100
- hle: 0.058,
1098
+ mmluPro: 0.664,
1099
+ gpqa: 0.491,
1100
+ hle: 0.036,
1101
1101
 
1102
1102
  // Capabilities
1103
1103
  contextWindow: 8192,
@@ -1105,18 +1105,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1105
1105
  supportsVision: false,
1106
1106
 
1107
1107
  // Metadata
1108
- lastUpdated: "2026-06-01",
1109
- originalModel: "Exaone 4.0 1.2B (Non-reasoning)",
1108
+ lastUpdated: "2026-07-01",
1109
+ originalModel: "Hermes 4 - Llama-3.1 70B (Non-reasoning)",
1110
1110
  },
1111
- "mimo-v2-flash-non-reasoning": {
1111
+ "hermes-4---llama-3.1-405b-non-reasoning": {
1112
1112
  // AA specific benchmarks
1113
- codingIndex: 25.8,
1114
- mathIndex: 67.7,
1113
+ codingIndex: undefined,
1114
+ mathIndex: 15.3,
1115
1115
 
1116
1116
  // Academic benchmarks
1117
- mmluPro: 0.744,
1118
- gpqa: 0.656,
1119
- hle: 0.08,
1117
+ mmluPro: 0.729,
1118
+ gpqa: 0.536,
1119
+ hle: 0.042,
1120
1120
 
1121
1121
  // Capabilities
1122
1122
  contextWindow: 8192,
@@ -1124,18 +1124,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1124
1124
  supportsVision: false,
1125
1125
 
1126
1126
  // Metadata
1127
- lastUpdated: "2026-06-01",
1128
- originalModel: "MiMo-V2-Flash (Non-reasoning)",
1127
+ lastUpdated: "2026-07-01",
1128
+ originalModel: "Hermes 4 - Llama-3.1 405B (Non-reasoning)",
1129
1129
  },
1130
- "mimo-v2.5-pro-non-reasoning": {
1130
+ "deephermes-3---llama-3.1-8b-preview-non-reasoning": {
1131
1131
  // AA specific benchmarks
1132
- codingIndex: 36.8,
1132
+ codingIndex: undefined,
1133
1133
  mathIndex: undefined,
1134
1134
 
1135
1135
  // Academic benchmarks
1136
- mmluPro: undefined,
1137
- gpqa: 0.762,
1138
- hle: 0.133,
1136
+ mmluPro: 0.365,
1137
+ gpqa: 0.27,
1138
+ hle: 0.043,
1139
1139
 
1140
1140
  // Capabilities
1141
1141
  contextWindow: 8192,
@@ -1143,18 +1143,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1143
1143
  supportsVision: false,
1144
1144
 
1145
1145
  // Metadata
1146
- lastUpdated: "2026-06-01",
1147
- originalModel: "MiMo-V2.5-Pro (Non-reasoning)",
1146
+ lastUpdated: "2026-07-01",
1147
+ originalModel: "DeepHermes 3 - Llama-3.1 8B Preview (Non-reasoning)",
1148
1148
  },
1149
- "mimo-v2.5": {
1149
+ "hermes-4---llama-3.1-405b-reasoning": {
1150
1150
  // AA specific benchmarks
1151
- codingIndex: 42.1,
1152
- mathIndex: undefined,
1151
+ codingIndex: undefined,
1152
+ mathIndex: 69.7,
1153
1153
 
1154
1154
  // Academic benchmarks
1155
- mmluPro: undefined,
1156
- gpqa: 0.849,
1157
- hle: 0.252,
1155
+ mmluPro: 0.829,
1156
+ gpqa: 0.727,
1157
+ hle: 0.103,
1158
1158
 
1159
1159
  // Capabilities
1160
1160
  contextWindow: 8192,
@@ -1162,18 +1162,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1162
1162
  supportsVision: false,
1163
1163
 
1164
1164
  // Metadata
1165
- lastUpdated: "2026-06-01",
1166
- originalModel: "MiMo-V2.5",
1165
+ lastUpdated: "2026-07-01",
1166
+ originalModel: "Hermes 4 - Llama-3.1 405B (Reasoning)",
1167
1167
  },
1168
- "mimo-v2.5-pro": {
1168
+ "hermes-4---llama-3.1-70b-reasoning": {
1169
1169
  // AA specific benchmarks
1170
- codingIndex: 45.5,
1171
- mathIndex: undefined,
1170
+ codingIndex: undefined,
1171
+ mathIndex: 68.7,
1172
1172
 
1173
1173
  // Academic benchmarks
1174
- mmluPro: undefined,
1175
- gpqa: 0.866,
1176
- hle: 0.338,
1174
+ mmluPro: 0.811,
1175
+ gpqa: 0.699,
1176
+ hle: 0.079,
1177
1177
 
1178
1178
  // Capabilities
1179
1179
  contextWindow: 8192,
@@ -1181,18 +1181,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1181
1181
  supportsVision: false,
1182
1182
 
1183
1183
  // Metadata
1184
- lastUpdated: "2026-06-01",
1185
- originalModel: "MiMo-V2.5-Pro",
1184
+ lastUpdated: "2026-07-01",
1185
+ originalModel: "Hermes 4 - Llama-3.1 70B (Reasoning)",
1186
1186
  },
1187
- "mimo-v2-flash-feb-2026": {
1187
+ "deephermes-3---mistral-24b-preview-non-reasoning": {
1188
1188
  // AA specific benchmarks
1189
- codingIndex: 33.5,
1189
+ codingIndex: undefined,
1190
1190
  mathIndex: undefined,
1191
1191
 
1192
1192
  // Academic benchmarks
1193
- mmluPro: undefined,
1194
- gpqa: 0.835,
1195
- hle: 0.2,
1193
+ mmluPro: 0.58,
1194
+ gpqa: 0.382,
1195
+ hle: 0.039,
1196
1196
 
1197
1197
  // Capabilities
1198
1198
  contextWindow: 8192,
@@ -1200,18 +1200,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1200
1200
  supportsVision: false,
1201
1201
 
1202
1202
  // Metadata
1203
- lastUpdated: "2026-06-01",
1204
- originalModel: "MiMo-V2-Flash (Feb 2026)",
1203
+ lastUpdated: "2026-07-01",
1204
+ originalModel: "DeepHermes 3 - Mistral 24B Preview (Non-reasoning)",
1205
1205
  },
1206
- "mimo-v2-omni-0327": {
1206
+ "exaone-4.5-33b": {
1207
1207
  // AA specific benchmarks
1208
- codingIndex: 36.9,
1208
+ codingIndex: undefined,
1209
1209
  mathIndex: undefined,
1210
1210
 
1211
1211
  // Academic benchmarks
1212
1212
  mmluPro: undefined,
1213
- gpqa: 0.855,
1214
- hle: 0.204,
1213
+ gpqa: 0.794,
1214
+ hle: 0.116,
1215
1215
 
1216
1216
  // Capabilities
1217
1217
  contextWindow: 8192,
@@ -1219,18 +1219,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1219
1219
  supportsVision: false,
1220
1220
 
1221
1221
  // Metadata
1222
- lastUpdated: "2026-06-01",
1223
- originalModel: "MiMo-V2-Omni-0327",
1222
+ lastUpdated: "2026-07-01",
1223
+ originalModel: "EXAONE 4.5 33B",
1224
1224
  },
1225
- "mimo-v2-omni": {
1225
+ "exaone-4.0-1.2b-reasoning": {
1226
1226
  // AA specific benchmarks
1227
- codingIndex: 35.5,
1228
- mathIndex: undefined,
1227
+ codingIndex: undefined,
1228
+ mathIndex: 50.3,
1229
1229
 
1230
1230
  // Academic benchmarks
1231
- mmluPro: undefined,
1232
- gpqa: 0.828,
1233
- hle: 0.199,
1231
+ mmluPro: 0.588,
1232
+ gpqa: 0.515,
1233
+ hle: 0.058,
1234
1234
 
1235
1235
  // Capabilities
1236
1236
  contextWindow: 8192,
@@ -1238,18 +1238,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1238
1238
  supportsVision: false,
1239
1239
 
1240
1240
  // Metadata
1241
- lastUpdated: "2026-06-01",
1242
- originalModel: "MiMo-V2-Omni",
1241
+ lastUpdated: "2026-07-01",
1242
+ originalModel: "Exaone 4.0 1.2B (Reasoning)",
1243
1243
  },
1244
- "ernie-4.5-300b-a47b": {
1244
+ "exaone-4.0-32b-non-reasoning": {
1245
1245
  // AA specific benchmarks
1246
- codingIndex: 14.5,
1247
- mathIndex: 41.3,
1246
+ codingIndex: undefined,
1247
+ mathIndex: 39.3,
1248
1248
 
1249
1249
  // Academic benchmarks
1250
- mmluPro: 0.776,
1251
- gpqa: 0.811,
1252
- hle: 0.035,
1250
+ mmluPro: 0.768,
1251
+ gpqa: 0.628,
1252
+ hle: 0.049,
1253
1253
 
1254
1254
  // Capabilities
1255
1255
  contextWindow: 8192,
@@ -1257,18 +1257,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1257
1257
  supportsVision: false,
1258
1258
 
1259
1259
  // Metadata
1260
- lastUpdated: "2026-06-01",
1261
- originalModel: "ERNIE 4.5 300B A47B",
1260
+ lastUpdated: "2026-07-01",
1261
+ originalModel: "EXAONE 4.0 32B (Non-reasoning)",
1262
1262
  },
1263
- "ernie-5.0-thinking-preview": {
1263
+ "exaone-4.0-32b-reasoning": {
1264
1264
  // AA specific benchmarks
1265
- codingIndex: 29.2,
1266
- mathIndex: 85,
1265
+ codingIndex: undefined,
1266
+ mathIndex: 80,
1267
1267
 
1268
1268
  // Academic benchmarks
1269
- mmluPro: 0.83,
1270
- gpqa: 0.777,
1271
- hle: 0.127,
1269
+ mmluPro: 0.818,
1270
+ gpqa: 0.739,
1271
+ hle: 0.105,
1272
1272
 
1273
1273
  // Capabilities
1274
1274
  contextWindow: 8192,
@@ -1276,18 +1276,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1276
1276
  supportsVision: false,
1277
1277
 
1278
1278
  // Metadata
1279
- lastUpdated: "2026-06-01",
1280
- originalModel: "ERNIE 5.0 Thinking Preview",
1279
+ lastUpdated: "2026-07-01",
1280
+ originalModel: "EXAONE 4.0 32B (Reasoning)",
1281
1281
  },
1282
- "sarvam-30b-high": {
1282
+ "k-exaone-non-reasoning": {
1283
1283
  // AA specific benchmarks
1284
- codingIndex: 7.9,
1285
- mathIndex: undefined,
1284
+ codingIndex: undefined,
1285
+ mathIndex: 44,
1286
1286
 
1287
1287
  // Academic benchmarks
1288
- mmluPro: undefined,
1289
- gpqa: 0.633,
1290
- hle: 0.07,
1288
+ mmluPro: 0.81,
1289
+ gpqa: 0.695,
1290
+ hle: 0.054,
1291
1291
 
1292
1292
  // Capabilities
1293
1293
  contextWindow: 8192,
@@ -1295,18 +1295,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1295
1295
  supportsVision: false,
1296
1296
 
1297
1297
  // Metadata
1298
- lastUpdated: "2026-06-01",
1299
- originalModel: "Sarvam 30B (high)",
1298
+ lastUpdated: "2026-07-01",
1299
+ originalModel: "K-EXAONE (Non-reasoning)",
1300
1300
  },
1301
- "sarvam-105b-high": {
1301
+ "k-exaone-reasoning": {
1302
1302
  // AA specific benchmarks
1303
- codingIndex: 9.8,
1304
- mathIndex: undefined,
1303
+ codingIndex: undefined,
1304
+ mathIndex: 90.3,
1305
1305
 
1306
1306
  // Academic benchmarks
1307
- mmluPro: undefined,
1308
- gpqa: 0.738,
1309
- hle: 0.101,
1307
+ mmluPro: 0.838,
1308
+ gpqa: 0.783,
1309
+ hle: 0.131,
1310
1310
 
1311
1311
  // Capabilities
1312
1312
  contextWindow: 8192,
@@ -1314,18 +1314,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1314
1314
  supportsVision: false,
1315
1315
 
1316
1316
  // Metadata
1317
- lastUpdated: "2026-06-01",
1318
- originalModel: "Sarvam 105B (high)",
1317
+ lastUpdated: "2026-07-01",
1318
+ originalModel: "K-EXAONE (Reasoning)",
1319
1319
  },
1320
- "qwen-chat-14b": {
1320
+ "exaone-4.0-1.2b-non-reasoning": {
1321
1321
  // AA specific benchmarks
1322
1322
  codingIndex: undefined,
1323
- mathIndex: undefined,
1323
+ mathIndex: 24,
1324
1324
 
1325
1325
  // Academic benchmarks
1326
- mmluPro: undefined,
1327
- gpqa: undefined,
1328
- hle: undefined,
1326
+ mmluPro: 0.5,
1327
+ gpqa: 0.424,
1328
+ hle: 0.058,
1329
1329
 
1330
1330
  // Capabilities
1331
1331
  contextWindow: 8192,
@@ -1333,18 +1333,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1333
1333
  supportsVision: false,
1334
1334
 
1335
1335
  // Metadata
1336
- lastUpdated: "2026-06-01",
1337
- originalModel: "Qwen Chat 14B",
1336
+ lastUpdated: "2026-07-01",
1337
+ originalModel: "Exaone 4.0 1.2B (Non-reasoning)",
1338
1338
  },
1339
- "hy3-preview-reasoning": {
1339
+ "mimo-v2.5-pro": {
1340
1340
  // AA specific benchmarks
1341
- codingIndex: 36.5,
1341
+ codingIndex: 60.2,
1342
1342
  mathIndex: undefined,
1343
1343
 
1344
1344
  // Academic benchmarks
1345
1345
  mmluPro: undefined,
1346
- gpqa: 0.867,
1347
- hle: 0.255,
1346
+ gpqa: 0.866,
1347
+ hle: 0.338,
1348
1348
 
1349
1349
  // Capabilities
1350
1350
  contextWindow: 8192,
@@ -1352,18 +1352,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1352
1352
  supportsVision: false,
1353
1353
 
1354
1354
  // Metadata
1355
- lastUpdated: "2026-06-01",
1356
- originalModel: "Hy3-preview (Reasoning)",
1355
+ lastUpdated: "2026-07-01",
1356
+ originalModel: "MiMo-V2.5-Pro",
1357
1357
  },
1358
- "hy3-preview-non-reasoning": {
1358
+ "mimo-v2-flash-non-reasoning": {
1359
1359
  // AA specific benchmarks
1360
- codingIndex: 34.3,
1361
- mathIndex: undefined,
1360
+ codingIndex: 49.8,
1361
+ mathIndex: 67.7,
1362
1362
 
1363
1363
  // Academic benchmarks
1364
- mmluPro: undefined,
1365
- gpqa: 0.732,
1366
- hle: 0.063,
1364
+ mmluPro: 0.744,
1365
+ gpqa: 0.656,
1366
+ hle: 0.08,
1367
1367
 
1368
1368
  // Capabilities
1369
1369
  contextWindow: 8192,
@@ -1371,18 +1371,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1371
1371
  supportsVision: false,
1372
1372
 
1373
1373
  // Metadata
1374
- lastUpdated: "2026-06-01",
1375
- originalModel: "Hy3-preview (Non-reasoning)",
1374
+ lastUpdated: "2026-07-01",
1375
+ originalModel: "MiMo-V2-Flash (Non-reasoning)",
1376
1376
  },
1377
- "kat-coder-pro-v2": {
1377
+ "mimo-v2.5-pro-non-reasoning": {
1378
1378
  // AA specific benchmarks
1379
- codingIndex: 45.6,
1379
+ codingIndex: undefined,
1380
1380
  mathIndex: undefined,
1381
1381
 
1382
1382
  // Academic benchmarks
1383
1383
  mmluPro: undefined,
1384
- gpqa: 0.855,
1385
- hle: 0.16,
1384
+ gpqa: 0.762,
1385
+ hle: 0.133,
1386
1386
 
1387
1387
  // Capabilities
1388
1388
  contextWindow: 8192,
@@ -1390,18 +1390,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1390
1390
  supportsVision: false,
1391
1391
 
1392
1392
  // Metadata
1393
- lastUpdated: "2026-06-01",
1394
- originalModel: "KAT Coder Pro V2",
1393
+ lastUpdated: "2026-07-01",
1394
+ originalModel: "MiMo-V2.5-Pro (Non-reasoning)",
1395
1395
  },
1396
- "kat-coder-pro-v1": {
1396
+ "mimo-v2.5": {
1397
1397
  // AA specific benchmarks
1398
- codingIndex: 18.3,
1399
- mathIndex: 94.7,
1398
+ codingIndex: undefined,
1399
+ mathIndex: undefined,
1400
1400
 
1401
1401
  // Academic benchmarks
1402
- mmluPro: 0.813,
1403
- gpqa: 0.764,
1404
- hle: 0.334,
1402
+ mmluPro: undefined,
1403
+ gpqa: 0.849,
1404
+ hle: 0.252,
1405
1405
 
1406
1406
  // Capabilities
1407
1407
  contextWindow: 8192,
@@ -1409,18 +1409,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1409
1409
  supportsVision: false,
1410
1410
 
1411
1411
  // Metadata
1412
- lastUpdated: "2026-06-01",
1413
- originalModel: "KAT-Coder-Pro V1",
1412
+ lastUpdated: "2026-07-01",
1413
+ originalModel: "MiMo-V2.5",
1414
1414
  },
1415
- "intellect-3": {
1415
+ "mimo-v2-flash-feb-2026": {
1416
1416
  // AA specific benchmarks
1417
- codingIndex: 19.1,
1418
- mathIndex: 88,
1417
+ codingIndex: undefined,
1418
+ mathIndex: undefined,
1419
1419
 
1420
1420
  // Academic benchmarks
1421
- mmluPro: 0.822,
1422
- gpqa: 0.761,
1423
- hle: 0.121,
1421
+ mmluPro: undefined,
1422
+ gpqa: 0.835,
1423
+ hle: 0.2,
1424
1424
 
1425
1425
  // Capabilities
1426
1426
  contextWindow: 8192,
@@ -1428,18 +1428,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1428
1428
  supportsVision: false,
1429
1429
 
1430
1430
  // Metadata
1431
- lastUpdated: "2026-06-01",
1432
- originalModel: "INTELLECT-3",
1431
+ lastUpdated: "2026-07-01",
1432
+ originalModel: "MiMo-V2-Flash (Feb 2026)",
1433
1433
  },
1434
- "motif-2-12.7b-reasoning": {
1434
+ "mimo-v2-omni": {
1435
1435
  // AA specific benchmarks
1436
- codingIndex: 11.9,
1437
- mathIndex: 80.3,
1436
+ codingIndex: undefined,
1437
+ mathIndex: undefined,
1438
1438
 
1439
1439
  // Academic benchmarks
1440
- mmluPro: 0.796,
1441
- gpqa: 0.695,
1442
- hle: 0.082,
1440
+ mmluPro: undefined,
1441
+ gpqa: 0.828,
1442
+ hle: 0.199,
1443
1443
 
1444
1444
  // Capabilities
1445
1445
  contextWindow: 8192,
@@ -1447,18 +1447,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1447
1447
  supportsVision: false,
1448
1448
 
1449
1449
  // Metadata
1450
- lastUpdated: "2026-06-01",
1451
- originalModel: "Motif-2-12.7B-Reasoning",
1450
+ lastUpdated: "2026-07-01",
1451
+ originalModel: "MiMo-V2-Omni",
1452
1452
  },
1453
- "k2-think-v2": {
1453
+ "mimo-v2-omni-0327": {
1454
1454
  // AA specific benchmarks
1455
- codingIndex: 15.5,
1455
+ codingIndex: undefined,
1456
1456
  mathIndex: undefined,
1457
1457
 
1458
1458
  // Academic benchmarks
1459
1459
  mmluPro: undefined,
1460
- gpqa: 0.713,
1461
- hle: 0.095,
1460
+ gpqa: 0.855,
1461
+ hle: 0.204,
1462
1462
 
1463
1463
  // Capabilities
1464
1464
  contextWindow: 8192,
@@ -1466,18 +1466,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1466
1466
  supportsVision: false,
1467
1467
 
1468
1468
  // Metadata
1469
- lastUpdated: "2026-06-01",
1470
- originalModel: "K2 Think V2",
1469
+ lastUpdated: "2026-07-01",
1470
+ originalModel: "MiMo-V2-Omni-0327",
1471
1471
  },
1472
- "k2-v2-high": {
1472
+ "ernie-5.0-thinking-preview": {
1473
1473
  // AA specific benchmarks
1474
- codingIndex: 16.1,
1475
- mathIndex: 78.3,
1474
+ codingIndex: undefined,
1475
+ mathIndex: 85,
1476
1476
 
1477
1477
  // Academic benchmarks
1478
- mmluPro: 0.786,
1479
- gpqa: 0.681,
1480
- hle: 0.098,
1478
+ mmluPro: 0.83,
1479
+ gpqa: 0.777,
1480
+ hle: 0.127,
1481
1481
 
1482
1482
  // Capabilities
1483
1483
  contextWindow: 8192,
@@ -1485,18 +1485,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1485
1485
  supportsVision: false,
1486
1486
 
1487
1487
  // Metadata
1488
- lastUpdated: "2026-06-01",
1489
- originalModel: "K2-V2 (high)",
1488
+ lastUpdated: "2026-07-01",
1489
+ originalModel: "ERNIE 5.0 Thinking Preview",
1490
1490
  },
1491
- "k2-v2-low": {
1491
+ "ernie-4.5-300b-a47b": {
1492
1492
  // AA specific benchmarks
1493
- codingIndex: 10.5,
1494
- mathIndex: 35.3,
1493
+ codingIndex: undefined,
1494
+ mathIndex: 41.3,
1495
1495
 
1496
1496
  // Academic benchmarks
1497
- mmluPro: 0.713,
1498
- gpqa: 0.541,
1499
- hle: 0.039,
1497
+ mmluPro: 0.776,
1498
+ gpqa: 0.811,
1499
+ hle: 0.035,
1500
1500
 
1501
1501
  // Capabilities
1502
1502
  contextWindow: 8192,
@@ -1504,18 +1504,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1504
1504
  supportsVision: false,
1505
1505
 
1506
1506
  // Metadata
1507
- lastUpdated: "2026-06-01",
1508
- originalModel: "K2-V2 (low)",
1507
+ lastUpdated: "2026-07-01",
1508
+ originalModel: "ERNIE 4.5 300B A47B",
1509
1509
  },
1510
- "k2-v2-medium": {
1510
+ "sarvam-30b-high": {
1511
1511
  // AA specific benchmarks
1512
- codingIndex: 14,
1513
- mathIndex: 64.7,
1512
+ codingIndex: undefined,
1513
+ mathIndex: undefined,
1514
1514
 
1515
1515
  // Academic benchmarks
1516
- mmluPro: 0.761,
1517
- gpqa: 0.598,
1518
- hle: 0.044,
1516
+ mmluPro: undefined,
1517
+ gpqa: 0.633,
1518
+ hle: 0.07,
1519
1519
 
1520
1520
  // Capabilities
1521
1521
  contextWindow: 8192,
@@ -1523,18 +1523,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1523
1523
  supportsVision: false,
1524
1524
 
1525
1525
  // Metadata
1526
- lastUpdated: "2026-06-01",
1527
- originalModel: "K2-V2 (medium)",
1526
+ lastUpdated: "2026-07-01",
1527
+ originalModel: "Sarvam 30B (high)",
1528
1528
  },
1529
- "mi-dm-k-2.5-pro": {
1529
+ "sarvam-105b-high": {
1530
1530
  // AA specific benchmarks
1531
- codingIndex: 12.6,
1532
- mathIndex: 76.7,
1531
+ codingIndex: undefined,
1532
+ mathIndex: undefined,
1533
1533
 
1534
1534
  // Academic benchmarks
1535
- mmluPro: 0.809,
1536
- gpqa: 0.701,
1537
- hle: 0.077,
1535
+ mmluPro: undefined,
1536
+ gpqa: 0.738,
1537
+ hle: 0.101,
1538
1538
 
1539
1539
  // Capabilities
1540
1540
  contextWindow: 8192,
@@ -1542,18 +1542,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1542
1542
  supportsVision: false,
1543
1543
 
1544
1544
  // Metadata
1545
- lastUpdated: "2026-06-01",
1546
- originalModel: "Mi:dm K 2.5 Pro",
1545
+ lastUpdated: "2026-07-01",
1546
+ originalModel: "Sarvam 105B (high)",
1547
1547
  },
1548
- "hyperclova-x-seed-think-32b": {
1548
+ "hypernova-60b-2605": {
1549
1549
  // AA specific benchmarks
1550
- codingIndex: 17.5,
1551
- mathIndex: 59,
1550
+ codingIndex: undefined,
1551
+ mathIndex: undefined,
1552
1552
 
1553
1553
  // Academic benchmarks
1554
- mmluPro: 0.785,
1555
- gpqa: 0.615,
1556
- hle: 0.055,
1554
+ mmluPro: undefined,
1555
+ gpqa: 0.733,
1556
+ hle: 0.151,
1557
1557
 
1558
1558
  // Capabilities
1559
1559
  contextWindow: 8192,
@@ -1561,18 +1561,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1561
1561
  supportsVision: false,
1562
1562
 
1563
1563
  // Metadata
1564
- lastUpdated: "2026-06-01",
1565
- originalModel: "HyperCLOVA X SEED Think (32B)",
1564
+ lastUpdated: "2026-07-01",
1565
+ originalModel: "HyperNova 60B 2605",
1566
1566
  },
1567
- "longcat-flash-lite": {
1567
+ "hy3-preview-reasoning": {
1568
1568
  // AA specific benchmarks
1569
- codingIndex: 16.5,
1569
+ codingIndex: undefined,
1570
1570
  mathIndex: undefined,
1571
1571
 
1572
1572
  // Academic benchmarks
1573
1573
  mmluPro: undefined,
1574
- gpqa: 0.636,
1575
- hle: 0.06,
1574
+ gpqa: 0.867,
1575
+ hle: 0.255,
1576
1576
 
1577
1577
  // Capabilities
1578
1578
  contextWindow: 8192,
@@ -1580,18 +1580,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1580
1580
  supportsVision: false,
1581
1581
 
1582
1582
  // Metadata
1583
- lastUpdated: "2026-06-01",
1584
- originalModel: "LongCat Flash Lite",
1583
+ lastUpdated: "2026-07-01",
1584
+ originalModel: "Hy3-preview (Reasoning)",
1585
1585
  },
1586
- "tri-21b-think": {
1586
+ "hy3-preview-non-reasoning": {
1587
1587
  // AA specific benchmarks
1588
- codingIndex: 6.3,
1588
+ codingIndex: undefined,
1589
1589
  mathIndex: undefined,
1590
1590
 
1591
1591
  // Academic benchmarks
1592
1592
  mmluPro: undefined,
1593
- gpqa: 0.601,
1594
- hle: 0.061,
1593
+ gpqa: 0.732,
1594
+ hle: 0.063,
1595
1595
 
1596
1596
  // Capabilities
1597
1597
  contextWindow: 8192,
@@ -1599,18 +1599,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1599
1599
  supportsVision: false,
1600
1600
 
1601
1601
  // Metadata
1602
- lastUpdated: "2026-06-01",
1603
- originalModel: "Tri-21B-Think",
1602
+ lastUpdated: "2026-07-01",
1603
+ originalModel: "Hy3-preview (Non-reasoning)",
1604
1604
  },
1605
- "tri-21b-think-preview": {
1605
+ "kat-coder-pro-v1": {
1606
1606
  // AA specific benchmarks
1607
- codingIndex: 7.4,
1608
- mathIndex: undefined,
1607
+ codingIndex: 58.9,
1608
+ mathIndex: 94.7,
1609
1609
 
1610
1610
  // Academic benchmarks
1611
- mmluPro: undefined,
1612
- gpqa: 0.538,
1613
- hle: 0.057,
1611
+ mmluPro: 0.813,
1612
+ gpqa: 0.764,
1613
+ hle: 0.334,
1614
1614
 
1615
1615
  // Capabilities
1616
1616
  contextWindow: 8192,
@@ -1618,18 +1618,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1618
1618
  supportsVision: false,
1619
1619
 
1620
1620
  // Metadata
1621
- lastUpdated: "2026-06-01",
1622
- originalModel: "Tri-21B-think Preview",
1621
+ lastUpdated: "2026-07-01",
1622
+ originalModel: "KAT-Coder-Pro V1",
1623
1623
  },
1624
- "nanbeige4.1-3b": {
1624
+ "kat-coder-pro-v2": {
1625
1625
  // AA specific benchmarks
1626
- codingIndex: 8.9,
1626
+ codingIndex: undefined,
1627
1627
  mathIndex: undefined,
1628
1628
 
1629
1629
  // Academic benchmarks
1630
1630
  mmluPro: undefined,
1631
- gpqa: 0.849,
1632
- hle: 0.1,
1631
+ gpqa: 0.855,
1632
+ hle: 0.16,
1633
1633
 
1634
1634
  // Capabilities
1635
1635
  contextWindow: 8192,
@@ -1637,18 +1637,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1637
1637
  supportsVision: false,
1638
1638
 
1639
1639
  // Metadata
1640
- lastUpdated: "2026-06-01",
1641
- originalModel: "Nanbeige4.1-3B",
1640
+ lastUpdated: "2026-07-01",
1641
+ originalModel: "KAT Coder Pro V2",
1642
1642
  },
1643
- "apertus-8b-instruct": {
1643
+ "intellect-3": {
1644
1644
  // AA specific benchmarks
1645
- codingIndex: 1.4,
1646
- mathIndex: undefined,
1645
+ codingIndex: undefined,
1646
+ mathIndex: 88,
1647
1647
 
1648
1648
  // Academic benchmarks
1649
- mmluPro: undefined,
1650
- gpqa: 0.256,
1651
- hle: 0.05,
1649
+ mmluPro: 0.822,
1650
+ gpqa: 0.761,
1651
+ hle: 0.121,
1652
1652
 
1653
1653
  // Capabilities
1654
1654
  contextWindow: 8192,
@@ -1656,18 +1656,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1656
1656
  supportsVision: false,
1657
1657
 
1658
1658
  // Metadata
1659
- lastUpdated: "2026-06-01",
1660
- originalModel: "Apertus 8B Instruct",
1659
+ lastUpdated: "2026-07-01",
1660
+ originalModel: "INTELLECT-3",
1661
1661
  },
1662
- "apertus-70b-instruct": {
1662
+ "motif-2-12.7b-reasoning": {
1663
1663
  // AA specific benchmarks
1664
- codingIndex: 1.9,
1665
- mathIndex: undefined,
1664
+ codingIndex: undefined,
1665
+ mathIndex: 80.3,
1666
1666
 
1667
1667
  // Academic benchmarks
1668
- mmluPro: undefined,
1669
- gpqa: 0.272,
1670
- hle: 0.055,
1668
+ mmluPro: 0.796,
1669
+ gpqa: 0.695,
1670
+ hle: 0.082,
1671
1671
 
1672
1672
  // Capabilities
1673
1673
  contextWindow: 8192,
@@ -1675,18 +1675,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1675
1675
  supportsVision: false,
1676
1676
 
1677
1677
  // Metadata
1678
- lastUpdated: "2026-06-01",
1679
- originalModel: "Apertus 70B Instruct",
1678
+ lastUpdated: "2026-07-01",
1679
+ originalModel: "Motif-2-12.7B-Reasoning",
1680
1680
  },
1681
- "minicpm5-1b-non-reasoning": {
1681
+ "k2-v2-medium": {
1682
1682
  // AA specific benchmarks
1683
- codingIndex: 0.5,
1684
- mathIndex: undefined,
1683
+ codingIndex: undefined,
1684
+ mathIndex: 64.7,
1685
1685
 
1686
1686
  // Academic benchmarks
1687
- mmluPro: undefined,
1688
- gpqa: 0.269,
1689
- hle: 0.046,
1687
+ mmluPro: 0.761,
1688
+ gpqa: 0.598,
1689
+ hle: 0.044,
1690
1690
 
1691
1691
  // Capabilities
1692
1692
  contextWindow: 8192,
@@ -1694,18 +1694,18 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1694
1694
  supportsVision: false,
1695
1695
 
1696
1696
  // Metadata
1697
- lastUpdated: "2026-06-01",
1698
- originalModel: "MiniCPM5-1B (Non-reasoning)",
1697
+ lastUpdated: "2026-07-01",
1698
+ originalModel: "K2-V2 (medium)",
1699
1699
  },
1700
- "minicpm-v-4.6-1.3b": {
1700
+ "k2-v2-high": {
1701
1701
  // AA specific benchmarks
1702
- codingIndex: 0.7,
1703
- mathIndex: undefined,
1702
+ codingIndex: undefined,
1703
+ mathIndex: 78.3,
1704
1704
 
1705
1705
  // Academic benchmarks
1706
- mmluPro: undefined,
1707
- gpqa: 0.305,
1708
- hle: 0.049,
1706
+ mmluPro: 0.786,
1707
+ gpqa: 0.681,
1708
+ hle: 0.098,
1709
1709
 
1710
1710
  // Capabilities
1711
1711
  contextWindow: 8192,
@@ -1713,7 +1713,7 @@ export const BENCHMARKS_CHUNK_1: Record<string, HardcodedBenchmark> = {
1713
1713
  supportsVision: false,
1714
1714
 
1715
1715
  // Metadata
1716
- lastUpdated: "2026-06-01",
1717
- originalModel: "MiniCPM-V 4.6 1.3B",
1716
+ lastUpdated: "2026-07-01",
1717
+ originalModel: "K2-V2 (high)",
1718
1718
  },
1719
1719
  };