free-coding-models 0.1.39 → 0.1.41
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.
- package/README.md +14 -14
- package/bin/free-coding-models.js +53 -27
- package/lib/utils.js +6 -6
- package/package.json +1 -1
- package/sources.js +59 -57
package/README.md
CHANGED
|
@@ -199,23 +199,23 @@ free-coding-models
|
|
|
199
199
|
|
|
200
200
|
## 🤖 Coding Models
|
|
201
201
|
|
|
202
|
-
**44 coding models** across 8 tiers, ranked by [
|
|
203
|
-
|
|
204
|
-
| Tier |
|
|
205
|
-
|
|
206
|
-
| **S+** |
|
|
207
|
-
| **S**
|
|
208
|
-
| **A+**
|
|
209
|
-
| **A**
|
|
210
|
-
| **A-**
|
|
211
|
-
| **B+**
|
|
212
|
-
| **B**
|
|
213
|
-
| **C**
|
|
202
|
+
**44 coding models** across 8 tiers, ranked by [SWE-bench Verified](https://www.swebench.com) — the industry-standard benchmark measuring real GitHub issue resolution. Scores are self-reported by providers unless noted.
|
|
203
|
+
|
|
204
|
+
| Tier | SWE-bench | Models |
|
|
205
|
+
|------|-----------|--------|
|
|
206
|
+
| **S+** ≥70% | GLM 5 (77.8%), Kimi K2.5 (76.8%), Step 3.5 Flash (74.4%), MiniMax M2.1 (74.0%), GLM 4.7 (73.8%), DeepSeek V3.2 (73.1%), Devstral 2 (72.2%), Kimi K2 Thinking (71.3%), Qwen3 Coder 480B (70.6%), Qwen3 235B (70.0%) |
|
|
207
|
+
| **S** 60–70% | MiniMax M2 (69.4%), DeepSeek V3.1 Terminus (68.4%), Qwen3 80B Thinking (68.0%), Qwen3.5 400B (68.0%), Kimi K2 Instruct (65.8%), Qwen3 80B Instruct (65.0%), DeepSeek V3.1 (62.0%), Llama 4 Maverick (62.0%), GPT OSS 120B (60.0%) |
|
|
208
|
+
| **A+** 50–60% | Mistral Large 675B (58.0%), Nemotron Ultra 253B (56.0%), Colosseum 355B (52.0%), QwQ 32B (50.0%) |
|
|
209
|
+
| **A** 40–50% | Nemotron Super 49B (49.0%), Mistral Medium 3 (48.0%), Qwen2.5 Coder 32B (46.0%), Magistral Small (45.0%), Llama 4 Scout (44.0%), Llama 3.1 405B (44.0%), Nemotron Nano 30B (43.0%), R1 Distill 32B (43.9%), GPT OSS 20B (42.0%) |
|
|
210
|
+
| **A-** 35–40% | Llama 3.3 70B (39.5%), Seed OSS 36B (38.0%), R1 Distill 14B (37.7%), Stockmark 100B (36.0%) |
|
|
211
|
+
| **B+** 30–35% | Ministral 14B (34.0%), Mixtral 8x22B (32.0%), Granite 34B Code (30.0%) |
|
|
212
|
+
| **B** 20–30% | R1 Distill 8B (28.2%), R1 Distill 7B (22.6%) |
|
|
213
|
+
| **C** <20% | Gemma 2 9B (18.0%), Phi 4 Mini (14.0%), Phi 3.5 Mini (12.0%) |
|
|
214
214
|
|
|
215
215
|
### Tier scale
|
|
216
216
|
|
|
217
|
-
- **S+/S** —
|
|
218
|
-
- **A+/A** —
|
|
217
|
+
- **S+/S** — Elite frontier coders (≥60% SWE-bench), best for complex real-world tasks and refactors
|
|
218
|
+
- **A+/A** — Great alternatives, strong at most coding tasks
|
|
219
219
|
- **A-/B+** — Solid performers, good for targeted programming tasks
|
|
220
220
|
- **B/C** — Lightweight or older models, good for code completion on constrained infra
|
|
221
221
|
|
|
@@ -421,7 +421,7 @@ function renderTable(results, pendingPings, frame, cursor = null, sortColumn = '
|
|
|
421
421
|
// 📖 Column widths (generous spacing with margins)
|
|
422
422
|
const W_RANK = 6
|
|
423
423
|
const W_TIER = 6
|
|
424
|
-
const
|
|
424
|
+
const W_CTX = 6
|
|
425
425
|
const W_SOURCE = 14
|
|
426
426
|
const W_MODEL = 26
|
|
427
427
|
const W_SWE = 9
|
|
@@ -454,7 +454,7 @@ function renderTable(results, pendingPings, frame, cursor = null, sortColumn = '
|
|
|
454
454
|
const originH = 'Origin'
|
|
455
455
|
const modelH = 'Model'
|
|
456
456
|
const sweH = sortColumn === 'swe' ? dir + ' SWE%' : 'SWE%'
|
|
457
|
-
const
|
|
457
|
+
const ctxH = sortColumn === 'ctx' ? dir + ' CTX' : 'CTX'
|
|
458
458
|
const pingH = sortColumn === 'ping' ? dir + ' Latest Ping' : 'Latest Ping'
|
|
459
459
|
const avgH = sortColumn === 'avg' ? dir + ' Avg Ping' : 'Avg Ping'
|
|
460
460
|
const healthH = sortColumn === 'condition' ? dir + ' Health' : 'Health'
|
|
@@ -477,15 +477,15 @@ function renderTable(results, pendingPings, frame, cursor = null, sortColumn = '
|
|
|
477
477
|
const originH_c = sortColumn === 'origin' ? chalk.bold.cyan(originH.padEnd(W_SOURCE)) : colorFirst(originH, W_SOURCE)
|
|
478
478
|
const modelH_c = colorFirst(modelH, W_MODEL)
|
|
479
479
|
const sweH_c = sortColumn === 'swe' ? chalk.bold.cyan(sweH.padEnd(W_SWE)) : colorFirst(sweH, W_SWE)
|
|
480
|
-
const
|
|
480
|
+
const ctxH_c = sortColumn === 'ctx' ? chalk.bold.cyan(ctxH.padEnd(W_CTX)) : colorFirst(ctxH, W_CTX)
|
|
481
481
|
const pingH_c = sortColumn === 'ping' ? chalk.bold.cyan(pingH.padEnd(W_PING)) : colorFirst('Latest Ping', W_PING)
|
|
482
482
|
const avgH_c = sortColumn === 'avg' ? chalk.bold.cyan(avgH.padEnd(W_AVG)) : colorFirst('Avg Ping', W_AVG)
|
|
483
483
|
const healthH_c = sortColumn === 'condition' ? chalk.bold.cyan(healthH.padEnd(W_STATUS)) : colorFirst('Health', W_STATUS)
|
|
484
484
|
const verdictH_c = sortColumn === 'verdict' ? chalk.bold.cyan(verdictH.padEnd(W_VERDICT)) : colorFirst(verdictH, W_VERDICT)
|
|
485
485
|
const uptimeH_c = sortColumn === 'uptime' ? chalk.bold.cyan(uptimeH.padStart(W_UPTIME)) : colorFirst(uptimeH, W_UPTIME, chalk.green)
|
|
486
486
|
|
|
487
|
-
// 📖 Header with proper spacing (column order: Rank, Tier, SWE%,
|
|
488
|
-
lines.push(' ' + rankH_c + ' ' + tierH_c + ' ' + sweH_c + ' ' +
|
|
487
|
+
// 📖 Header with proper spacing (column order: Rank, Tier, SWE%, CTX, Model, Origin, Latest Ping, Avg Ping, Health, Verdict, Up%)
|
|
488
|
+
lines.push(' ' + rankH_c + ' ' + tierH_c + ' ' + sweH_c + ' ' + ctxH_c + ' ' + modelH_c + ' ' + originH_c + ' ' + pingH_c + ' ' + avgH_c + ' ' + healthH_c + ' ' + verdictH_c + ' ' + uptimeH_c)
|
|
489
489
|
|
|
490
490
|
// 📖 Separator line
|
|
491
491
|
lines.push(
|
|
@@ -493,7 +493,7 @@ function renderTable(results, pendingPings, frame, cursor = null, sortColumn = '
|
|
|
493
493
|
chalk.dim('─'.repeat(W_RANK)) + ' ' +
|
|
494
494
|
chalk.dim('─'.repeat(W_TIER)) + ' ' +
|
|
495
495
|
chalk.dim('─'.repeat(W_SWE)) + ' ' +
|
|
496
|
-
chalk.dim('─'.repeat(
|
|
496
|
+
chalk.dim('─'.repeat(W_CTX)) + ' ' +
|
|
497
497
|
'─'.repeat(W_MODEL) + ' ' +
|
|
498
498
|
'─'.repeat(W_SOURCE) + ' ' +
|
|
499
499
|
chalk.dim('─'.repeat(W_PING)) + ' ' +
|
|
@@ -529,12 +529,12 @@ function renderTable(results, pendingPings, frame, cursor = null, sortColumn = '
|
|
|
529
529
|
: chalk.dim(sweScore.padEnd(W_SWE))
|
|
530
530
|
|
|
531
531
|
// 📖 Context window column - colorized by size (larger = better)
|
|
532
|
-
const
|
|
533
|
-
const
|
|
534
|
-
? chalk.greenBright(
|
|
535
|
-
:
|
|
536
|
-
? chalk.cyan(
|
|
537
|
-
: chalk.dim(
|
|
532
|
+
const ctxRaw = r.ctx ?? '—'
|
|
533
|
+
const ctxCell = ctxRaw !== '—' && (ctxRaw.includes('128k') || ctxRaw.includes('200k') || ctxRaw.includes('1m'))
|
|
534
|
+
? chalk.greenBright(ctxRaw.padEnd(W_CTX))
|
|
535
|
+
: ctxRaw !== '—' && (ctxRaw.includes('32k') || ctxRaw.includes('64k'))
|
|
536
|
+
? chalk.cyan(ctxRaw.padEnd(W_CTX))
|
|
537
|
+
: chalk.dim(ctxRaw.padEnd(W_CTX))
|
|
538
538
|
|
|
539
539
|
// 📖 Latest ping - pings are objects: { ms, code }
|
|
540
540
|
// 📖 Only show response time for successful pings, "—" for errors (error code is in Status column)
|
|
@@ -640,8 +640,8 @@ function renderTable(results, pendingPings, frame, cursor = null, sortColumn = '
|
|
|
640
640
|
uptimeCell = chalk.red(uptimeStr.padStart(W_UPTIME))
|
|
641
641
|
}
|
|
642
642
|
|
|
643
|
-
// 📖 Build row with double space between columns (order: Rank, Tier, SWE%,
|
|
644
|
-
const row = ' ' + num + ' ' + tier + ' ' + sweCell + ' ' +
|
|
643
|
+
// 📖 Build row with double space between columns (order: Rank, Tier, SWE%, CTX, Model, Origin, Latest Ping, Avg Ping, Health, Verdict, Up%)
|
|
644
|
+
const row = ' ' + num + ' ' + tier + ' ' + sweCell + ' ' + ctxCell + ' ' + name + ' ' + source + ' ' + pingCell + ' ' + avgCell + ' ' + status + ' ' + speedCell + ' ' + uptimeCell
|
|
645
645
|
|
|
646
646
|
if (isCursor) {
|
|
647
647
|
lines.push(chalk.bgRgb(139, 0, 139)(row))
|
|
@@ -1105,8 +1105,8 @@ async function runFiableMode(apiKey) {
|
|
|
1105
1105
|
console.log(chalk.cyan(' ⚡ Analyzing models for reliability (10 seconds)...'))
|
|
1106
1106
|
console.log()
|
|
1107
1107
|
|
|
1108
|
-
let results = MODELS.map(([modelId, label, tier, sweScore,
|
|
1109
|
-
idx: i + 1, modelId, label, tier, sweScore,
|
|
1108
|
+
let results = MODELS.map(([modelId, label, tier, sweScore, ctx], i) => ({
|
|
1109
|
+
idx: i + 1, modelId, label, tier, sweScore, ctx,
|
|
1110
1110
|
status: 'pending',
|
|
1111
1111
|
pings: [],
|
|
1112
1112
|
httpCode: null,
|
|
@@ -1183,21 +1183,47 @@ async function main() {
|
|
|
1183
1183
|
}
|
|
1184
1184
|
}
|
|
1185
1185
|
|
|
1186
|
-
// 📖
|
|
1187
|
-
|
|
1188
|
-
|
|
1186
|
+
// 📖 Check for updates in the background
|
|
1187
|
+
let latestVersion = null
|
|
1188
|
+
try {
|
|
1189
|
+
latestVersion = await checkForUpdate()
|
|
1190
|
+
} catch {
|
|
1191
|
+
// Silently fail - don't block the app if npm registry is unreachable
|
|
1192
|
+
}
|
|
1189
1193
|
|
|
1190
1194
|
// 📖 Default mode: OpenCode CLI
|
|
1191
1195
|
let mode = 'opencode'
|
|
1192
1196
|
|
|
1193
|
-
// 📖
|
|
1194
|
-
|
|
1195
|
-
|
|
1196
|
-
|
|
1197
|
+
// 📖 Show update notification menu if a new version is available
|
|
1198
|
+
if (latestVersion) {
|
|
1199
|
+
const action = await promptUpdateNotification(latestVersion)
|
|
1200
|
+
if (action === 'update') {
|
|
1201
|
+
runUpdate(latestVersion)
|
|
1202
|
+
return // runUpdate will restart the process
|
|
1203
|
+
} else if (action === 'changelogs') {
|
|
1204
|
+
console.log()
|
|
1205
|
+
console.log(chalk.cyan(' Opening changelog in browser...'))
|
|
1206
|
+
console.log()
|
|
1207
|
+
const { execSync } = require('child_process')
|
|
1208
|
+
const changelogUrl = 'https://github.com/vava-nessa/free-coding-models/releases'
|
|
1209
|
+
try {
|
|
1210
|
+
if (isMac) {
|
|
1211
|
+
execSync(`open "${changelogUrl}"`, { stdio: 'ignore' })
|
|
1212
|
+
} else if (isWindows) {
|
|
1213
|
+
execSync(`start "" "${changelogUrl}"`, { stdio: 'ignore' })
|
|
1214
|
+
} else {
|
|
1215
|
+
execSync(`xdg-open "${changelogUrl}"`, { stdio: 'ignore' })
|
|
1216
|
+
}
|
|
1217
|
+
} catch {
|
|
1218
|
+
console.log(chalk.dim(` Could not open browser. Visit: ${changelogUrl}`))
|
|
1219
|
+
}
|
|
1220
|
+
}
|
|
1221
|
+
// If action is null (Continue without update) or changelogs, proceed to main app
|
|
1222
|
+
}
|
|
1197
1223
|
|
|
1198
1224
|
// 📖 Create results array with all models initially visible
|
|
1199
|
-
let results = MODELS.map(([modelId, label, tier, sweScore,
|
|
1200
|
-
idx: i + 1, modelId, label, tier, sweScore,
|
|
1225
|
+
let results = MODELS.map(([modelId, label, tier, sweScore, ctx], i) => ({
|
|
1226
|
+
idx: i + 1, modelId, label, tier, sweScore, ctx,
|
|
1201
1227
|
status: 'pending',
|
|
1202
1228
|
pings: [], // 📖 All ping results (ms or 'TIMEOUT')
|
|
1203
1229
|
httpCode: null,
|
|
@@ -1306,10 +1332,10 @@ async function main() {
|
|
|
1306
1332
|
const onKeyPress = async (str, key) => {
|
|
1307
1333
|
if (!key) return
|
|
1308
1334
|
|
|
1309
|
-
// 📖 Sorting keys: R=rank, T=tier, O=origin, M=model, L=latest ping, A=avg ping, S=SWE-bench,
|
|
1335
|
+
// 📖 Sorting keys: R=rank, T=tier, O=origin, M=model, L=latest ping, A=avg ping, S=SWE-bench, N=context, H=health, V=verdict, U=uptime
|
|
1310
1336
|
const sortKeys = {
|
|
1311
1337
|
'r': 'rank', 't': 'tier', 'o': 'origin', 'm': 'model',
|
|
1312
|
-
'l': 'ping', 'a': 'avg', 's': 'swe', '
|
|
1338
|
+
'l': 'ping', 'a': 'avg', 's': 'swe', 'n': 'ctx', 'h': 'condition', 'v': 'verdict', 'u': 'uptime'
|
|
1313
1339
|
}
|
|
1314
1340
|
|
|
1315
1341
|
if (sortKeys[key.name]) {
|
package/lib/utils.js
CHANGED
|
@@ -136,7 +136,7 @@ export const getUptime = (r) => {
|
|
|
136
136
|
// - 'ping' (L key) — last ping latency (only successful ones count)
|
|
137
137
|
// - 'avg' (A key) — average latency across all successful pings
|
|
138
138
|
// - 'swe' (S key) — SWE-bench score (higher is better)
|
|
139
|
-
// - '
|
|
139
|
+
// - 'ctx' (N key) — context window size (larger is better)
|
|
140
140
|
// - 'condition' (H key) — health status (alphabetical)
|
|
141
141
|
// - 'verdict' (V key) — verdict order (Perfect → Pending)
|
|
142
142
|
// - 'uptime' (U key) — uptime percentage
|
|
@@ -185,12 +185,12 @@ export const sortResults = (results, sortColumn, sortDirection) => {
|
|
|
185
185
|
cmp = parseSwe(a.sweScore) - parseSwe(b.sweScore)
|
|
186
186
|
break
|
|
187
187
|
}
|
|
188
|
-
case '
|
|
188
|
+
case 'ctx': {
|
|
189
189
|
// 📖 Sort by context window size — larger is better
|
|
190
190
|
// 📖 Parse strings like "128k", "32k", "1m" into numeric tokens
|
|
191
|
-
const
|
|
192
|
-
if (!
|
|
193
|
-
const str =
|
|
191
|
+
const parseCtx = (ctx) => {
|
|
192
|
+
if (!ctx || ctx === '—') return 0
|
|
193
|
+
const str = ctx.toLowerCase()
|
|
194
194
|
// 📖 Handle millions (1m = 1000k)
|
|
195
195
|
if (str.includes('m')) {
|
|
196
196
|
const num = parseFloat(str.replace('m', ''))
|
|
@@ -203,7 +203,7 @@ export const sortResults = (results, sortColumn, sortDirection) => {
|
|
|
203
203
|
}
|
|
204
204
|
return 0
|
|
205
205
|
}
|
|
206
|
-
cmp =
|
|
206
|
+
cmp = parseCtx(a.ctx) - parseCtx(b.ctx)
|
|
207
207
|
break
|
|
208
208
|
}
|
|
209
209
|
case 'condition':
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "free-coding-models",
|
|
3
|
-
"version": "0.1.
|
|
3
|
+
"version": "0.1.41",
|
|
4
4
|
"description": "Find the fastest coding LLM models in seconds — ping free models from multiple providers, pick the best one for OpenCode, Cursor, or any AI coding assistant.",
|
|
5
5
|
"keywords": [
|
|
6
6
|
"nvidia",
|
package/sources.js
CHANGED
|
@@ -4,13 +4,13 @@
|
|
|
4
4
|
*
|
|
5
5
|
* @details
|
|
6
6
|
* This file contains all model definitions organized by provider/source.
|
|
7
|
-
* Each source has its own models array with [model_id, display_label, tier, swe_score,
|
|
7
|
+
* Each source has its own models array with [model_id, display_label, tier, swe_score, ctx].
|
|
8
8
|
* - model_id: The model identifier for API calls
|
|
9
9
|
* - display_label: Human-friendly name for display
|
|
10
10
|
* - tier: Performance tier (S+, S, A+, A, A-, B+, B, C)
|
|
11
|
-
* - swe_score: SWE-bench Verified score percentage
|
|
12
|
-
* -
|
|
13
|
-
*
|
|
11
|
+
* - swe_score: SWE-bench Verified score percentage (self-reported by model provider)
|
|
12
|
+
* - ctx: Context window size in tokens (e.g., "128k", "32k")
|
|
13
|
+
*
|
|
14
14
|
* Add new sources here to support additional providers beyond NIM.
|
|
15
15
|
*
|
|
16
16
|
* 🎯 Tier scale (based on SWE-bench Verified):
|
|
@@ -18,12 +18,14 @@
|
|
|
18
18
|
* - S: 60-70% (excellent)
|
|
19
19
|
* - A+: 50-60% (great)
|
|
20
20
|
* - A: 40-50% (good)
|
|
21
|
-
* - A-: 35-
|
|
22
|
-
* - B+: 30-
|
|
21
|
+
* - A-: 35-40% (decent)
|
|
22
|
+
* - B+: 30-35% (average)
|
|
23
23
|
* - B: 20-30% (below average)
|
|
24
24
|
* - C: <20% (lightweight/edge)
|
|
25
25
|
*
|
|
26
|
-
* 📖 Source: https://www.swebench.com
|
|
26
|
+
* 📖 Source: https://www.swebench.com — scores are self-reported unless noted
|
|
27
|
+
* 📖 Secondary: https://swe-rebench.com (independent evals, scores are lower)
|
|
28
|
+
* 📖 Leaderboard tracker: https://www.marc0.dev/en/leaderboard
|
|
27
29
|
*
|
|
28
30
|
* @exports Object containing all sources and their models
|
|
29
31
|
*/
|
|
@@ -31,57 +33,57 @@
|
|
|
31
33
|
// 📖 NIM source - https://build.nvidia.com
|
|
32
34
|
export const nvidiaNim = [
|
|
33
35
|
// ── S+ tier — SWE-bench Verified ≥70% ──
|
|
34
|
-
['deepseek-ai/deepseek-v3.1', 'DeepSeek V3.1', 'S+', '49.2%', '128k'],
|
|
35
|
-
['deepseek-ai/deepseek-v3.1-terminus', 'DeepSeek V3.1 Term', 'S+', '49.2%', '128k'],
|
|
36
36
|
['deepseek-ai/deepseek-v3.2', 'DeepSeek V3.2', 'S+', '73.1%', '128k'],
|
|
37
37
|
['moonshotai/kimi-k2.5', 'Kimi K2.5', 'S+', '76.8%', '128k'],
|
|
38
|
-
['
|
|
39
|
-
['
|
|
40
|
-
['
|
|
41
|
-
|
|
42
|
-
['
|
|
43
|
-
['
|
|
44
|
-
['qwen/qwen3
|
|
45
|
-
['
|
|
46
|
-
|
|
47
|
-
['
|
|
48
|
-
['
|
|
49
|
-
|
|
50
|
-
['
|
|
51
|
-
['
|
|
52
|
-
['qwen/qwen3-
|
|
53
|
-
['
|
|
54
|
-
['
|
|
55
|
-
['
|
|
56
|
-
// ── A tier — SWE-bench Verified
|
|
57
|
-
['
|
|
58
|
-
['mistralai/mistral-
|
|
59
|
-
['
|
|
60
|
-
['
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
['
|
|
64
|
-
['nvidia/llama-3.3-nemotron-super-49b-v1.5', 'Nemotron Super 49B', 'A
|
|
65
|
-
['meta/llama-4-scout-17b-16e-instruct', 'Llama 4 Scout', 'A
|
|
38
|
+
['z-ai/glm5', 'GLM 5', 'S+', '77.8%', '128k'],
|
|
39
|
+
['z-ai/glm4.7', 'GLM 4.7', 'S+', '73.8%', '200k'],
|
|
40
|
+
['moonshotai/kimi-k2-thinking', 'Kimi K2 Thinking', 'S+', '71.3%', '256k'],
|
|
41
|
+
['minimaxai/minimax-m2.1', 'MiniMax M2.1', 'S+', '74.0%', '200k'],
|
|
42
|
+
['stepfun-ai/step-3.5-flash', 'Step 3.5 Flash', 'S+', '74.4%', '256k'],
|
|
43
|
+
['qwen/qwen3-coder-480b-a35b-instruct', 'Qwen3 Coder 480B', 'S+', '70.6%', '256k'],
|
|
44
|
+
['qwen/qwen3-235b-a22b', 'Qwen3 235B', 'S+', '70.0%', '128k'],
|
|
45
|
+
['mistralai/devstral-2-123b-instruct-2512', 'Devstral 2 123B', 'S+', '72.2%', '256k'],
|
|
46
|
+
// ── S tier — SWE-bench Verified 60–70% ──
|
|
47
|
+
['deepseek-ai/deepseek-v3.1-terminus', 'DeepSeek V3.1 Term', 'S', '68.4%', '128k'],
|
|
48
|
+
['moonshotai/kimi-k2-instruct', 'Kimi K2 Instruct', 'S', '65.8%', '128k'],
|
|
49
|
+
['minimaxai/minimax-m2', 'MiniMax M2', 'S', '69.4%', '128k'],
|
|
50
|
+
['qwen/qwen3-next-80b-a3b-thinking', 'Qwen3 80B Thinking', 'S', '68.0%', '128k'],
|
|
51
|
+
['qwen/qwen3-next-80b-a3b-instruct', 'Qwen3 80B Instruct', 'S', '65.0%', '128k'],
|
|
52
|
+
['qwen/qwen3.5-397b-a17b', 'Qwen3.5 400B VLM', 'S', '68.0%', '128k'],
|
|
53
|
+
['openai/gpt-oss-120b', 'GPT OSS 120B', 'S', '60.0%', '128k'],
|
|
54
|
+
['meta/llama-4-maverick-17b-128e-instruct', 'Llama 4 Maverick', 'S', '62.0%', '1M'],
|
|
55
|
+
['deepseek-ai/deepseek-v3.1', 'DeepSeek V3.1', 'S', '62.0%', '128k'],
|
|
56
|
+
// ── A+ tier — SWE-bench Verified 50–60% ──
|
|
57
|
+
['nvidia/llama-3.1-nemotron-ultra-253b-v1', 'Nemotron Ultra 253B', 'A+', '56.0%', '128k'],
|
|
58
|
+
['mistralai/mistral-large-3-675b-instruct-2512', 'Mistral Large 675B', 'A+', '58.0%', '256k'],
|
|
59
|
+
['qwen/qwq-32b', 'QwQ 32B', 'A+', '50.0%', '131k'],
|
|
60
|
+
['igenius/colosseum_355b_instruct_16k', 'Colosseum 355B', 'A+', '52.0%', '16k'],
|
|
61
|
+
// ── A tier — SWE-bench Verified 40–50% ──
|
|
62
|
+
['mistralai/mistral-medium-3-instruct', 'Mistral Medium 3', 'A', '48.0%', '128k'],
|
|
63
|
+
['mistralai/magistral-small-2506', 'Magistral Small', 'A', '45.0%', '32k'],
|
|
64
|
+
['nvidia/llama-3.3-nemotron-super-49b-v1.5', 'Nemotron Super 49B', 'A', '49.0%', '128k'],
|
|
65
|
+
['meta/llama-4-scout-17b-16e-instruct', 'Llama 4 Scout', 'A', '44.0%', '10M'],
|
|
66
|
+
['nvidia/nemotron-3-nano-30b-a3b', 'Nemotron Nano 30B', 'A', '43.0%', '128k'],
|
|
67
|
+
['deepseek-ai/deepseek-r1-distill-qwen-32b', 'R1 Distill 32B', 'A', '43.9%', '128k'],
|
|
68
|
+
['openai/gpt-oss-20b', 'GPT OSS 20B', 'A', '42.0%', '128k'],
|
|
69
|
+
['qwen/qwen2.5-coder-32b-instruct', 'Qwen2.5 Coder 32B', 'A', '46.0%', '32k'],
|
|
70
|
+
['meta/llama-3.1-405b-instruct', 'Llama 3.1 405B', 'A', '44.0%', '128k'],
|
|
71
|
+
// ── A- tier — SWE-bench Verified 35–40% ──
|
|
72
|
+
['meta/llama-3.3-70b-instruct', 'Llama 3.3 70B', 'A-', '39.5%', '128k'],
|
|
66
73
|
['deepseek-ai/deepseek-r1-distill-qwen-14b', 'R1 Distill 14B', 'A-', '37.7%', '64k'],
|
|
67
|
-
['
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
['
|
|
71
|
-
['
|
|
72
|
-
['
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
['
|
|
76
|
-
|
|
77
|
-
['
|
|
78
|
-
['
|
|
79
|
-
['
|
|
80
|
-
// ── C tier — SWE-bench Verified <25% or lightweight edge models ──
|
|
81
|
-
['deepseek-ai/deepseek-r1-distill-qwen-7b', 'R1 Distill 7B', 'C', '22.6%', '32k'],
|
|
82
|
-
['google/gemma-2-9b-it', 'Gemma 2 9B', 'C', '18.0%', '8k'],
|
|
83
|
-
['microsoft/phi-3.5-mini-instruct', 'Phi 3.5 Mini', 'C', '12.0%', '128k'],
|
|
84
|
-
['microsoft/phi-4-mini-instruct', 'Phi 4 Mini', 'C', '14.0%', '128k'],
|
|
74
|
+
['bytedance/seed-oss-36b-instruct', 'Seed OSS 36B', 'A-', '38.0%', '32k'],
|
|
75
|
+
['stockmark/stockmark-2-100b-instruct', 'Stockmark 100B', 'A-', '36.0%', '32k'],
|
|
76
|
+
// ── B+ tier — SWE-bench Verified 30–35% ──
|
|
77
|
+
['mistralai/mixtral-8x22b-instruct-v0.1', 'Mixtral 8x22B', 'B+', '32.0%', '64k'],
|
|
78
|
+
['mistralai/ministral-14b-instruct-2512', 'Ministral 14B', 'B+', '34.0%', '32k'],
|
|
79
|
+
['ibm/granite-34b-code-instruct', 'Granite 34B Code', 'B+', '30.0%', '32k'],
|
|
80
|
+
// ── B tier — SWE-bench Verified 20–30% ──
|
|
81
|
+
['deepseek-ai/deepseek-r1-distill-llama-8b', 'R1 Distill 8B', 'B', '28.2%', '32k'],
|
|
82
|
+
['deepseek-ai/deepseek-r1-distill-qwen-7b', 'R1 Distill 7B', 'B', '22.6%', '32k'],
|
|
83
|
+
// ── C tier — SWE-bench Verified <20% or lightweight edge models ──
|
|
84
|
+
['google/gemma-2-9b-it', 'Gemma 2 9B', 'C', '18.0%', '8k'],
|
|
85
|
+
['microsoft/phi-3.5-mini-instruct', 'Phi 3.5 Mini', 'C', '12.0%', '128k'],
|
|
86
|
+
['microsoft/phi-4-mini-instruct', 'Phi 4 Mini', 'C', '14.0%', '128k'],
|
|
85
87
|
]
|
|
86
88
|
|
|
87
89
|
// 📖 All sources combined - used by the main script
|
|
@@ -95,7 +97,7 @@ export const sources = {
|
|
|
95
97
|
// 📖 Flatten all models from all sources for backward compatibility
|
|
96
98
|
export const MODELS = []
|
|
97
99
|
for (const [sourceKey, sourceData] of Object.entries(sources)) {
|
|
98
|
-
for (const [modelId, label, tier, sweScore,
|
|
99
|
-
MODELS.push([modelId, label, tier, sweScore,
|
|
100
|
+
for (const [modelId, label, tier, sweScore, ctx] of sourceData.models) {
|
|
101
|
+
MODELS.push([modelId, label, tier, sweScore, ctx])
|
|
100
102
|
}
|
|
101
103
|
}
|