@blockrun/clawrouter 0.12.245 → 0.12.247

package/dist/index.js

@@ -35385,6 +35385,21 @@ function applyPromotions(tierConfigs, promotions, profile, now2 = /* @__PURE__ *
   }
   return result;
 }
+function applyUnavailableModels(tierConfigs, unavailableModels) {
+  if (!unavailableModels || unavailableModels.length === 0) return tierConfigs;
+  const dead = new Set(unavailableModels);
+  let result = tierConfigs;
+  for (const tier of Object.keys(tierConfigs)) {
+    const config = tierConfigs[tier];
+    const alive = [config.primary, ...config.fallback].filter((model) => !dead.has(model));
+    if (alive.length === 0 || alive[0] === config.primary && alive.length === config.fallback.length + 1) {
+      continue;
+    }
+    if (result === tierConfigs) result = { ...tierConfigs };
+    result[tier] = { primary: alive[0], fallback: alive.slice(1) };
+  }
+  return result;
+}
 function getStrategy(name) {
   const strategy = registry.get(name);
   if (!strategy) {
@@ -36082,6 +36097,7 @@ ${value.slice(-(scanLimit - prefixLength))}`;
           profile = useAgenticTiers ? "agentic" : "auto";
         }
         tierConfigs = applyPromotions(tierConfigs, config.promotions, profile, options.now);
+        tierConfigs = applyUnavailableModels(tierConfigs, options.unavailableModels);
         const agenticScoreValue = ruleResult.agenticScore;
         if (estimatedTokens > config.overrides.maxTokensForceComplex) {
           const decision2 = selectModel(
@@ -36210,18 +36226,6 @@ ${value.slice(-(scanLimit - prefixLength))}`;
         supportsTools: false,
         supportsVision: false
       },
-      "free/gpt-oss-120b": {
-        contextWindow: 128e3,
-        maxOutputTokens: 16384,
-        supportsTools: false,
-        supportsVision: false
-      },
-      "free/gpt-oss-20b": {
-        contextWindow: 128e3,
-        maxOutputTokens: 16384,
-        supportsTools: false,
-        supportsVision: false
-      },
       "free/seed-oss-36b": {
         contextWindow: 131072,
         maxOutputTokens: 16384,
@@ -36294,6 +36298,18 @@ ${value.slice(-(scanLimit - prefixLength))}`;
         supportsTools: true,
         supportsVision: true
       },
+      "nvidia/nemotron-nano-9b-v2": {
+        contextWindow: 131072,
+        maxOutputTokens: 16384,
+        supportsTools: false,
+        supportsVision: false
+      },
+      "nvidia/step-3.7-flash": {
+        contextWindow: 131072,
+        maxOutputTokens: 16384,
+        supportsTools: false,
+        supportsVision: false
+      },
       "openai/gpt-4.1": {
         contextWindow: 128e3,
         maxOutputTokens: 16384,
@@ -36767,9 +36783,12 @@ ${value.slice(-(scanLimit - prefixLength))}`;
         const targetTier = (features.taskType === "reasoning_mcq" || features.taskType === "reasoning_math") && (base3.tier === "SIMPLE" || base3.tier === "MEDIUM") ? "REASONING" : base3.tier;
         const tierConfig = tierConfigs[targetTier];
         const configuredCandidates = tierConfig ? getFallbackChain(targetTier, tierConfigs) : [];
+        const unavailable = new Set(options.unavailableModels ?? []);
         const chain3 = [
           .../* @__PURE__ */ new Set([...configuredCandidates, ...evidenceCandidates(features.taskType)])
-        ].filter((model2) => typeof model2 === "string" && model2.length > 0);
+        ].filter(
+          (model2) => typeof model2 === "string" && model2.length > 0 && !unavailable.has(model2)
+        );
         const eligible = chain3.filter(
           (model2) => isEligible(model2, features, maxOutputTokens, options)
         );
@@ -36846,13 +36865,7 @@ ${value.slice(-(scanLimit - prefixLength))}`;
           ...eligibleCandidates.filter(
             (model2) => !scoredModels.includes(model2) && !webResearchFallbackOrder.includes(model2)
           )
-        ] : features.taskType === "tool_agent" || features.taskType === "tool_agent_parallel" && features.agentDomain !== "other" ? [
-          ...scoredModels,
-          ...eligibleCandidates.filter((model2) => !scoredModels.includes(model2))
-        ] : [
-          ...scoredModels,
-          ...eligibleCandidates.filter((model2) => !scoredModels.includes(model2))
-        ];
+        ] : [...scoredModels, ...eligibleCandidates.filter((model2) => !scoredModels.includes(model2))];
         const model = ranked[0] ?? base3.model;
         const selectedTierConfigs = {
           ...tierConfigs,
@@ -37964,8 +37977,8 @@ ${value.slice(-(scanLimit - prefixLength))}`;
             // $0.20/$1.25, 1M context
             "xai/grok-4-fast-non-reasoning",
             // 1,143ms, $0.20/$0.50 — fast fallback
-            "free/gpt-oss-120b"
-            // 1,252ms, FREE fallback (hidden from /v1/models but direct calls work)
+            "nvidia/step-3.7-flash"
+            // FREE backstop — new NVIDIA free tier (gpt-oss-120b now 400s; probed 2026-08-21)
           ]
         },
         MEDIUM: {
@@ -38038,14 +38051,15 @@ ${value.slice(-(scanLimit - prefixLength))}`;
       // Eco tier configs - absolute cheapest (blockrun/eco)
       ecoTiers: {
         SIMPLE: {
-          primary: "free/gpt-oss-120b",
-          // FREE! $0.00/$0.00 — heavy user default
+          primary: "nvidia/step-3.7-flash",
+          // FREE! $0.00/$0.00 — new NVIDIA free tier flagship
           fallback: [
-            "free/gpt-oss-20b",
-            // FREE — smaller, faster
-            // deepseek-v4-flash and seed-oss-36b sat here until NVIDIA EOL'd them
-            // (410; 2026-08-12 and 2026-08-03 respectively). gpt-oss-120b/20b already
-            // head this chain, so the rungs are dropped, not retargeted.
+            "nvidia/nemotron-nano-9b-v2",
+            // FREE — compact + fast (~0.7s), high-volume light tasks
+            // This head keeps rotting with NVIDIA's free hosting: deepseek-v4-flash
+            // (410, 2026-08-12), seed-oss-36b (410, 2026-08-03), then gpt-oss-120b/20b
+            // (400 Unknown model, probed 2026-08-21). Each retirement retargets the
+            // two free rungs to the current free tier; the paid rungs below never move.
             "google/gemini-3.1-flash-lite",
             // $0.25/$1.50 — newest flash-lite
             "openai/gpt-5.4-nano",
@@ -38162,8 +38176,8 @@ ${value.slice(-(scanLimit - prefixLength))}`;
             "openai/gpt-5.3-codex",
             "deepseek/deepseek-chat",
             // Cheap, reliable
-            "free/gpt-oss-120b"
-            // NVIDIA free ultimate backstop (was seed-oss-36b; EOL'd 2026-08-03)
+            "nvidia/step-3.7-flash"
+            // NVIDIA free ultimate backstop (was gpt-oss-120b; 400s since ~2026-08)
           ]
         },
         REASONING: {
@@ -38252,8 +38266,8 @@ ${value.slice(-(scanLimit - prefixLength))}`;
             // Previous flagship — 6,213ms, reliable
             "deepseek/deepseek-chat",
             // 1,431ms — cheap, reliable
-            "free/gpt-oss-120b"
-            // NVIDIA free ultimate backstop (was seed-oss-36b; EOL'd 2026-08-03)
+            "nvidia/step-3.7-flash"
+            // NVIDIA free ultimate backstop (was gpt-oss-120b; 400s since ~2026-08)
           ]
         },
         REASONING: {
@@ -57791,6 +57805,14 @@ Then run: npx @blockrun/clawrouter`
     solanaAddress = await getSolanaAddress(derived.solanaPrivateKeyBytes);
   } catch {
   }
+  let solanaDefaultSaved = false;
+  if (solanaAddress) {
+    try {
+      await savePaymentChain("solana");
+      solanaDefaultSaved = true;
+    } catch {
+    }
+  }
   console.log(`[ClawRouter]`);
   console.log(`[ClawRouter] \u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550`);
   console.log(`[ClawRouter]   NEW WALLET GENERATED \u2014 BACK UP YOUR KEY NOW`);
@@ -57803,6 +57825,12 @@ Then run: npx @blockrun/clawrouter`
   console.log(`[ClawRouter]   Mnemonic       : ${MNEMONIC_FILE}`);
   console.log(`[ClawRouter]`);
   console.log(`[ClawRouter]   Both EVM (Base) and Solana wallets are ready.`);
+  if (solanaDefaultSaved) {
+    console.log(
+      `[ClawRouter]   Default payment chain: Solana \u2014 fund the Solana address above with USDC.`
+    );
+    console.log(`[ClawRouter]   To switch to Base, run in OpenClaw: /wallet base`);
+  }
   console.log(`[ClawRouter]   To back up, run in OpenClaw:`);
   console.log(`[ClawRouter]     /wallet export`);
   console.log(`[ClawRouter]`);
@@ -90608,25 +90636,7 @@ async function proxyRequest(req, res, apiBase, payFetch, options, routerOpts, de
         const modelMatch = imageArgs.match(/--model\s+(\S+)/);
         if (modelMatch) {
           const raw = modelMatch[1];
-          const IMAGE_MODEL_ALIASES = {
-            // dall-e-3 delisted upstream 2026-05-25 — legacy aliases route to
-            // the OpenAI successor instead of a guaranteed 400.
-            "dall-e-3": "openai/gpt-image-2",
-            dalle3: "openai/gpt-image-2",
-            dalle: "openai/gpt-image-2",
-            "gpt-image": "openai/gpt-image-1",
-            "gpt-image-1": "openai/gpt-image-1",
-            "gpt-image-2": "openai/gpt-image-2",
-            seedream: "bytedance/seedream-5-pro",
-            banana: "google/nano-banana",
-            "nano-banana": "google/nano-banana",
-            "banana-pro": "google/nano-banana-pro",
-            "nano-banana-pro": "google/nano-banana-pro",
-            "grok-imagine": "xai/grok-imagine-image",
-            "grok-imagine-pro": "xai/grok-imagine-image-pro",
-            cogview: "zai/cogview-4"
-          };
-          imageModel = IMAGE_MODEL_ALIASES[raw] ?? raw;
+          imageModel = Object.prototype.hasOwnProperty.call(IMAGE_MODEL_ALIASES, raw) ? IMAGE_MODEL_ALIASES[raw] : raw;
           imagePrompt = imagePrompt.replace(/--model\s+\S+/, "").trim();
         }
         const sizeMatch = imageArgs.match(/--size\s+(\d+x\d+)/);
@@ -90644,11 +90654,13 @@ async function proxyRequest(req, res, apiBase, payFetch, options, routerOpts, de
             "",
             "Models:",
             "  nano-banana       Google Gemini Flash \u2014 $0.05/image",
+            "  banana-2          Google Nano Banana 2 \u2014 $0.09/image",
             "  banana-pro        Google Gemini Pro \u2014 $0.10/image (up to 4K)",
             "  gpt-image         OpenAI GPT Image 1 \u2014 $0.02/image",
             "  gpt-image-2       OpenAI GPT Image 2 \u2014 $0.06/image",
             "  seedream          ByteDance Seedream 5 Pro \u2014 $0.045/image",
             "  grok-imagine      xAI Grok Imagine \u2014 $0.02/image",
+            "  grok-imagine-pro  xAI Grok Imagine Pro \u2014 $0.07/image",
             "  cogview           Zhipu CogView-4 \u2014 $0.015/image",
             "",
             "Examples:",
@@ -90711,14 +90723,68 @@ async function proxyRequest(req, res, apiBase, payFetch, options, routerOpts, de
             size: imageSize,
             n: 1
           });
+          const imagegenAbort = new AbortController();
+          res.on("close", () => {
+            if (!res.writableEnded) imagegenAbort.abort();
+          });
           const imageResponse = await payFetch(imageUpstreamUrl, {
             method: "POST",
             headers: { "content-type": "application/json", "user-agent": USER_AGENT },
-            body: imageBody
+            body: imageBody,
+            signal: imagegenAbort.signal
           });
-          const imageResult = await imageResponse.json();
+          let imageResult = await imageResponse.json();
+          let imageFailure;
+          if (imageResponse.ok && imageResult.poll_url && imageResult.id && !imageResult.data?.length) {
+            const apiOrigin = new URL(apiBase).origin;
+            const pollUrl = imageResult.poll_url.startsWith("http") ? imageResult.poll_url : `${apiOrigin}${imageResult.poll_url}`;
+            console.log(
+              `[ClawRouter] /imagegen job submitted (id=${imageResult.id}), polling upstream \u2014 typical 30\u2013120s...`
+            );
+            const pollDeadline = Date.now() + 3e5;
+            await new Promise((r2) => setTimeout(r2, 2e3));
+            let completed2 = false;
+            while (Date.now() < pollDeadline) {
+              if (imagegenAbort.signal.aborted) {
+                console.log(
+                  `[ClawRouter] Chat client disconnected \u2014 abandoning image poll (id=${imageResult.id})`
+                );
+                return;
+              }
+              const pollResp = await payFetch(pollUrl, {
+                method: "GET",
+                headers: { "user-agent": USER_AGENT },
+                signal: imagegenAbort.signal
+              });
+              const pollText = await pollResp.text();
+              let pollBody = {};
+              try {
+                pollBody = JSON.parse(pollText);
+              } catch {
+                imageFailure = `Non-JSON poll response (${pollResp.status}): ${pollText.slice(0, 200)}`;
+                break;
+              }
+              if (pollResp.status === 202 || pollBody.status === "queued" || pollBody.status === "in_progress") {
+                await new Promise((r2) => setTimeout(r2, 3e3));
+                continue;
+              }
+              if (pollBody.status === "failed" || !pollResp.ok) {
+                imageFailure = typeof pollBody.error === "string" ? pollBody.error : pollBody.error?.message ?? `Upstream image generation failed (HTTP ${pollResp.status})`;
+                break;
+              }
+              imageResult = pollBody;
+              completed2 = true;
+              break;
+            }
+            if (!completed2 && !imageFailure) {
+              imageFailure = "Image generation timed out after 5 minutes \u2014 the job may still complete upstream.";
+            }
+          }
           let responseText;
-          if (!imageResponse.ok || imageResult.error) {
+          if (imageFailure) {
+            responseText = `Image generation failed: ${imageFailure}`;
+            console.log(`[ClawRouter] /imagegen error: ${imageFailure}`);
+          } else if (!imageResponse.ok || imageResult.error) {
             const errMsg = typeof imageResult.error === "string" ? imageResult.error : imageResult.error?.message ?? `HTTP ${imageResponse.status}`;
             responseText = `Image generation failed: ${errMsg}`;
             console.log(`[ClawRouter] /imagegen error: ${errMsg}`);
@@ -90738,9 +90804,7 @@ async function proxyRequest(req, res, apiBase, payFetch, options, routerOpts, de
                       console.error(
                         `[ClawRouter] /imagegen: failed to upload data URI: ${uploadErr instanceof Error ? uploadErr.message : String(uploadErr)}`
                       );
-                      lines.push(
-                        "Image generated but upload failed. Try again or use --model dall-e-3."
-                      );
+                      lines.push("Image generated but upload failed. Try again.");
                     }
                   } else {
                     lines.push(img.url);
@@ -92524,7 +92588,7 @@ data: [DONE]
     });
   }
 }
-var paymentStore, BLOCKRUN_API, BLOCKRUN_SOLANA_API, IMAGE_DIR, AUDIO_DIR, VIDEO_DIR, AUTO_MODEL, ROUTING_PROFILES, FREE_MODELS, FREE_MODEL, MAX_MESSAGES, CONTEXT_LIMIT_KB, HEARTBEAT_INTERVAL_MS, BALANCE_CHECK_TIMEOUT_MS, DEFAULT_REQUEST_TIMEOUT_MS, PER_MODEL_TIMEOUT_MS, REASONING_MODEL_TIMEOUT_MS, REASONING_MODEL_IDS, MAX_FALLBACK_ATTEMPTS, HEALTH_CHECK_TIMEOUT_MS, RATE_LIMIT_COOLDOWN_MS, OVERLOAD_COOLDOWN_MS, PORT_RETRY_ATTEMPTS, PORT_RETRY_DELAY_MS, MODEL_BODY_READ_TIMEOUT_MS, ERROR_BODY_READ_TIMEOUT_MS, rateLimitedModels, overloadedModels, perProviderErrors, BALANCE_CHECK_BUFFER, BALANCE_PREFLIGHT_OUTPUT_TOKEN_CAP, PROVIDER_ERROR_PATTERNS, DEGRADED_RESPONSE_PATTERNS, DEGRADED_LOOP_PATTERNS, VALID_ROLES, ROLE_MAPPINGS, VALID_TOOL_ID_PATTERN, KIMI_BLOCK_RE, KIMI_TOKEN_RE, THINKING_TAG_RE, THINKING_BLOCK_RE, BLOCKRUN_MODEL_BY_ID, IMAGE_PRICING, VIDEO_PRICING, PHONE_PRICING;
+var paymentStore, BLOCKRUN_API, BLOCKRUN_SOLANA_API, IMAGE_DIR, AUDIO_DIR, VIDEO_DIR, AUTO_MODEL, ROUTING_PROFILES, FREE_MODELS, FREE_MODEL, MAX_MESSAGES, CONTEXT_LIMIT_KB, HEARTBEAT_INTERVAL_MS, BALANCE_CHECK_TIMEOUT_MS, DEFAULT_REQUEST_TIMEOUT_MS, PER_MODEL_TIMEOUT_MS, REASONING_MODEL_TIMEOUT_MS, REASONING_MODEL_IDS, MAX_FALLBACK_ATTEMPTS, HEALTH_CHECK_TIMEOUT_MS, RATE_LIMIT_COOLDOWN_MS, OVERLOAD_COOLDOWN_MS, PORT_RETRY_ATTEMPTS, PORT_RETRY_DELAY_MS, MODEL_BODY_READ_TIMEOUT_MS, ERROR_BODY_READ_TIMEOUT_MS, rateLimitedModels, overloadedModels, perProviderErrors, BALANCE_CHECK_BUFFER, BALANCE_PREFLIGHT_OUTPUT_TOKEN_CAP, PROVIDER_ERROR_PATTERNS, DEGRADED_RESPONSE_PATTERNS, DEGRADED_LOOP_PATTERNS, VALID_ROLES, ROLE_MAPPINGS, VALID_TOOL_ID_PATTERN, KIMI_BLOCK_RE, KIMI_TOKEN_RE, THINKING_TAG_RE, THINKING_BLOCK_RE, BLOCKRUN_MODEL_BY_ID, IMAGE_PRICING, IMAGE_MODEL_IDS, IMAGE_MODEL_SIZES, IMAGE_MODEL_ALIASES, VIDEO_PRICING, PHONE_PRICING;
 var init_proxy = __esm({
   "src/proxy.ts"() {
     "use strict";
@@ -92669,9 +92733,20 @@ var init_proxy = __esm({
       },
       "bytedance/seedream-5-pro": {
         default: 0.045,
-        sizes: { "1024x1024": 0.045, "2048x1024": 0.045, "2048x2048": 0.09 }
+        // Full size list live-probed 2026-08-23 (gateway 402 quotes, base price):
+        // the 2K portrait/landscape tiers bill at the 2048x2048 rate.
+        sizes: {
+          "1024x1024": 0.045,
+          "1280x720": 0.045,
+          "2048x1024": 0.045,
+          "2048x2048": 0.09,
+          "2304x1728": 0.09,
+          "1728x2304": 0.09,
+          "2848x1600": 0.09,
+          "1600x2848": 0.09
+        }
       },
-      "google/nano-banana": { default: 0.05 },
+      "google/nano-banana": { default: 0.05, sizes: { "1024x1024": 0.05 } },
       // Nano Banana 2 (Gemini 3.1 Flash imagegen, blockrun #329): $0.09/image at 1K.
       "google/nano-banana-2": { default: 0.09, sizes: { "1024x1024": 0.09 } },
       "google/nano-banana-pro": {
@@ -92692,6 +92767,30 @@ var init_proxy = __esm({
         }
       }
     };
+    IMAGE_MODEL_IDS = Object.freeze(Object.keys(IMAGE_PRICING));
+    IMAGE_MODEL_SIZES = Object.freeze(
+      [...new Set(Object.values(IMAGE_PRICING).flatMap((p) => Object.keys(p.sizes ?? {})))].sort()
+    );
+    IMAGE_MODEL_ALIASES = Object.freeze({
+      // dall-e-3 delisted upstream 2026-05-25 — legacy aliases route to
+      // the OpenAI successor instead of a guaranteed 400.
+      "dall-e-3": "openai/gpt-image-2",
+      dalle3: "openai/gpt-image-2",
+      dalle: "openai/gpt-image-2",
+      "gpt-image": "openai/gpt-image-1",
+      "gpt-image-1": "openai/gpt-image-1",
+      "gpt-image-2": "openai/gpt-image-2",
+      seedream: "bytedance/seedream-5-pro",
+      banana: "google/nano-banana",
+      "nano-banana": "google/nano-banana",
+      "banana-2": "google/nano-banana-2",
+      "nano-banana-2": "google/nano-banana-2",
+      "banana-pro": "google/nano-banana-pro",
+      "nano-banana-pro": "google/nano-banana-pro",
+      "grok-imagine": "xai/grok-imagine-image",
+      "grok-imagine-pro": "xai/grok-imagine-image-pro",
+      cogview: "zai/cogview-4"
+    });
     VIDEO_PRICING = {
       "xai/grok-imagine-video": { pricePerSecond: 0.05, defaultDurationSeconds: 8 },
       // Seedance rates re-synced 2026-08-08 from blockrun's models.ts after the
@@ -93465,8 +93564,8 @@ var init_registry = __esm({
         name: "Image Generation",
         partner: "BlockRun",
         category: "Image & Video",
-        shortDescription: "8 image models (DALL-E, Flux, Grok, ...)",
-        description: "Generate an image from a text prompt. Models available: google/nano-banana (default), google/nano-banana-pro (up to 4K), openai/gpt-image-1, openai/dall-e-3, black-forest/flux-1.1-pro, xai/grok-imagine-image, xai/grok-imagine-image-pro, zai/cogview-4. Returns a local http://localhost:8402/images/<file>.png URL.",
+        shortDescription: "9 image models (Nano Banana, GPT Image, ...)",
+        description: "Generate an image from a text prompt. Models available: google/nano-banana (default), google/nano-banana-2, google/nano-banana-pro (up to 4K), openai/gpt-image-1, openai/gpt-image-2, bytedance/seedream-5-pro, xai/grok-imagine-image, xai/grok-imagine-image-pro, zai/cogview-4. Returns a local http://localhost:8402/images/<file>.png URL.",
         proxyPath: "/images/generations",
         method: "POST",
         params: [
@@ -93479,7 +93578,7 @@ var init_registry = __esm({
           {
             name: "model",
             type: "string",
-            description: "Full model ID (e.g. 'google/nano-banana', 'openai/dall-e-3'). Default: google/nano-banana.",
+            description: "Full model ID (e.g. 'google/nano-banana', 'openai/gpt-image-2'). Default: google/nano-banana.",
             required: false
           },
           {
@@ -225847,12 +225946,23 @@ function buildImageGenerationProvider() {
     id: "blockrun",
     label: "BlockRun",
     defaultModel: "google/nano-banana",
+    // Must stay in sync with IMAGE_PRICING (proxy.ts). OpenClaw sends the
+    // picked id straight to /v1/images/generations, which forwards the body
+    // verbatim with no alias resolution, so a retired id here is a guaranteed
+    // upstream 400. dall-e-3 (delisted 2026-05-25) and flux-1.1-pro (no
+    // gateway entry) were dropped in v0.12.227 and had lingered here.
+    // Kept as a hand-written array (not spread from IMAGE_MODEL_IDS) because
+    // this is the picker's curated display order — default model first, which
+    // IMAGE_PRICING declaration order does not give — and because the
+    // lifecycle tests fully mock ./proxy.js. Set-equality is enforced by
+    // src/index.image-provider.test.ts, which pins the two lists together.
     models: [
       "google/nano-banana",
+      "google/nano-banana-2",
       "google/nano-banana-pro",
       "openai/gpt-image-1",
-      "openai/dall-e-3",
-      "black-forest/flux-1.1-pro",
+      "openai/gpt-image-2",
+      "bytedance/seedream-5-pro",
       "xai/grok-imagine-image",
       "xai/grok-imagine-image-pro",
       "zai/cogview-4"
@@ -225867,20 +225977,28 @@ function buildImageGenerationProvider() {
       // Only openai/gpt-image-1 supports edit server-side; OpenClaw's UI picks a
       // compatible model at edit time via /v1/images/image2image.
       edit: { enabled: true },
+      // Union of every size the gateway accepts across the models above,
+      // live-probed 2026-08-23 (the gateway validates size per-model BEFORE
+      // payment and 400s unknown ones — 1216x832 / 1792x1024 / 1024x1792 were
+      // accepted by no model and are gone with dall-e-3). Pinned against
+      // IMAGE_MODEL_SIZES (proxy.ts) by src/index.image-provider.test.ts.
       geometry: {
         sizes: [
           "512x512",
           "768x768",
           "768x1344",
           "1024x1024",
-          "1216x832",
+          "1024x1536",
+          "1280x720",
           "1344x768",
           "1440x1440",
           "1536x1024",
-          "1024x1536",
-          "1792x1024",
-          "1024x1792",
+          "1600x2848",
+          "1728x2304",
+          "2048x1024",
           "2048x2048",
+          "2304x1728",
+          "2848x1600",
           "4096x4096"
         ]
       }
@@ -226846,6 +226964,7 @@ export {
   SpendControl,
   VISIBLE_OPENCLAW_MODELS,
   blockrunProvider,
+  buildImageGenerationProvider,
   buildPartnerTools,
   buildProviderModels,
   calculateModelCost,

package/dist/router/index.js

@@ -363,6 +363,21 @@ function applyPromotions(tierConfigs, promotions, profile, now = /* @__PURE__ */
   }
   return result;
 }
+function applyUnavailableModels(tierConfigs, unavailableModels) {
+  if (!unavailableModels || unavailableModels.length === 0) return tierConfigs;
+  const dead = new Set(unavailableModels);
+  let result = tierConfigs;
+  for (const tier of Object.keys(tierConfigs)) {
+    const config = tierConfigs[tier];
+    const alive = [config.primary, ...config.fallback].filter((model) => !dead.has(model));
+    if (alive.length === 0 || alive[0] === config.primary && alive.length === config.fallback.length + 1) {
+      continue;
+    }
+    if (result === tierConfigs) result = { ...tierConfigs };
+    result[tier] = { primary: alive[0], fallback: alive.slice(1) };
+  }
+  return result;
+}
 var RulesStrategy = class {
   name = "rules";
   route(prompt, systemPrompt, maxOutputTokens, options) {
@@ -414,6 +429,7 @@ ${value.slice(-(scanLimit - prefixLength))}`;
       profile = useAgenticTiers ? "agentic" : "auto";
     }
     tierConfigs = applyPromotions(tierConfigs, config.promotions, profile, options.now);
+    tierConfigs = applyUnavailableModels(tierConfigs, options.unavailableModels);
     const agenticScoreValue = ruleResult.agenticScore;
     if (estimatedTokens > config.overrides.maxTokensForceComplex) {
       const decision2 = selectModel(
@@ -552,18 +568,6 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: false,
     supportsVision: false
   },
-  "free/gpt-oss-120b": {
-    contextWindow: 128e3,
-    maxOutputTokens: 16384,
-    supportsTools: false,
-    supportsVision: false
-  },
-  "free/gpt-oss-20b": {
-    contextWindow: 128e3,
-    maxOutputTokens: 16384,
-    supportsTools: false,
-    supportsVision: false
-  },
   "free/seed-oss-36b": {
     contextWindow: 131072,
     maxOutputTokens: 16384,
@@ -636,6 +640,18 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "nvidia/nemotron-nano-9b-v2": {
+    contextWindow: 131072,
+    maxOutputTokens: 16384,
+    supportsTools: false,
+    supportsVision: false
+  },
+  "nvidia/step-3.7-flash": {
+    contextWindow: 131072,
+    maxOutputTokens: 16384,
+    supportsTools: false,
+    supportsVision: false
+  },
   "openai/gpt-4.1": {
     contextWindow: 128e3,
     maxOutputTokens: 16384,
@@ -1732,9 +1748,12 @@ var PortfolioStrategy = class {
     const targetTier = (features.taskType === "reasoning_mcq" || features.taskType === "reasoning_math") && (base.tier === "SIMPLE" || base.tier === "MEDIUM") ? "REASONING" : base.tier;
     const tierConfig = tierConfigs[targetTier];
     const configuredCandidates = tierConfig ? getFallbackChain(targetTier, tierConfigs) : [];
+    const unavailable = new Set(options.unavailableModels ?? []);
     const chain = [
       .../* @__PURE__ */ new Set([...configuredCandidates, ...evidenceCandidates(features.taskType)])
-    ].filter((model2) => typeof model2 === "string" && model2.length > 0);
+    ].filter(
+      (model2) => typeof model2 === "string" && model2.length > 0 && !unavailable.has(model2)
+    );
     const eligible = chain.filter(
       (model2) => isEligible(model2, features, maxOutputTokens, options)
     );
@@ -1811,13 +1830,7 @@ var PortfolioStrategy = class {
       ...eligibleCandidates.filter(
         (model2) => !scoredModels.includes(model2) && !webResearchFallbackOrder.includes(model2)
       )
-    ] : features.taskType === "tool_agent" || features.taskType === "tool_agent_parallel" && features.agentDomain !== "other" ? [
-      ...scoredModels,
-      ...eligibleCandidates.filter((model2) => !scoredModels.includes(model2))
-    ] : [
-      ...scoredModels,
-      ...eligibleCandidates.filter((model2) => !scoredModels.includes(model2))
-    ];
+    ] : [...scoredModels, ...eligibleCandidates.filter((model2) => !scoredModels.includes(model2))];
     const model = ranked[0] ?? base.model;
     const selectedTierConfigs = {
       ...tierConfigs,
@@ -2929,8 +2942,8 @@ var DEFAULT_ROUTING_CONFIG = {
         // $0.20/$1.25, 1M context
         "xai/grok-4-fast-non-reasoning",
         // 1,143ms, $0.20/$0.50 — fast fallback
-        "free/gpt-oss-120b"
-        // 1,252ms, FREE fallback (hidden from /v1/models but direct calls work)
+        "nvidia/step-3.7-flash"
+        // FREE backstop — new NVIDIA free tier (gpt-oss-120b now 400s; probed 2026-08-21)
       ]
     },
     MEDIUM: {
@@ -3003,14 +3016,15 @@ var DEFAULT_ROUTING_CONFIG = {
   // Eco tier configs - absolute cheapest (blockrun/eco)
   ecoTiers: {
     SIMPLE: {
-      primary: "free/gpt-oss-120b",
-      // FREE! $0.00/$0.00 — heavy user default
+      primary: "nvidia/step-3.7-flash",
+      // FREE! $0.00/$0.00 — new NVIDIA free tier flagship
       fallback: [
-        "free/gpt-oss-20b",
-        // FREE — smaller, faster
-        // deepseek-v4-flash and seed-oss-36b sat here until NVIDIA EOL'd them
-        // (410; 2026-08-12 and 2026-08-03 respectively). gpt-oss-120b/20b already
-        // head this chain, so the rungs are dropped, not retargeted.
+        "nvidia/nemotron-nano-9b-v2",
+        // FREE — compact + fast (~0.7s), high-volume light tasks
+        // This head keeps rotting with NVIDIA's free hosting: deepseek-v4-flash
+        // (410, 2026-08-12), seed-oss-36b (410, 2026-08-03), then gpt-oss-120b/20b
+        // (400 Unknown model, probed 2026-08-21). Each retirement retargets the
+        // two free rungs to the current free tier; the paid rungs below never move.
         "google/gemini-3.1-flash-lite",
         // $0.25/$1.50 — newest flash-lite
         "openai/gpt-5.4-nano",
@@ -3127,8 +3141,8 @@ var DEFAULT_ROUTING_CONFIG = {
         "openai/gpt-5.3-codex",
         "deepseek/deepseek-chat",
         // Cheap, reliable
-        "free/gpt-oss-120b"
-        // NVIDIA free ultimate backstop (was seed-oss-36b; EOL'd 2026-08-03)
+        "nvidia/step-3.7-flash"
+        // NVIDIA free ultimate backstop (was gpt-oss-120b; 400s since ~2026-08)
       ]
     },
     REASONING: {
@@ -3217,8 +3231,8 @@ var DEFAULT_ROUTING_CONFIG = {
         // Previous flagship — 6,213ms, reliable
         "deepseek/deepseek-chat",
         // 1,431ms — cheap, reliable
-        "free/gpt-oss-120b"
-        // NVIDIA free ultimate backstop (was seed-oss-36b; EOL'd 2026-08-03)
+        "nvidia/step-3.7-flash"
+        // NVIDIA free ultimate backstop (was gpt-oss-120b; 400s since ~2026-08)
       ]
     },
     REASONING: {
@@ -3275,6 +3289,7 @@ export {
   LIVE_MODEL_PROFILES,
   PortfolioStrategy,
   RulesStrategy,
+  applyUnavailableModels,
   calculateModelCost,
   classifyByRules,
   filterByExcludeList,

package/docs/anthropic-cost-savings.md

@@ -50,7 +50,7 @@ This is where you're paying for real value:
 
 ## The Solution: ClawRouter
 
-[ClawRouter](https://github.com/BlockRunAI/ClawRouter) is an open-source local proxy that sits between your app and <!-- br:models.chatVisible -->70<!-- /br:models.chatVisible --> AI models. It saves you money in three ways: **smart routing**, **token optimization**, and **response caching**.
+[ClawRouter](https://github.com/BlockRunAI/ClawRouter) is an open-source local proxy that sits between your app and <!-- br:models.chatVisible -->71<!-- /br:models.chatVisible --> AI models. It saves you money in three ways: **smart routing**, **token optimization**, and **response caching**.
 
 ```
 ┌─────────────┐     ┌──────────────────────────────┐     ┌──────────────────┐

package/docs/anthropic-third-party-harness-changes.md

@@ -31,7 +31,7 @@ This is exactly the problem [ClawRouter](https://github.com/BlockRunAI/ClawRoute
 
 ## ClawRouter: Smart Routing for Agents
 
-ClawRouter is an open-source local proxy that sits between your agent and <!-- br:models.chatVisible -->70<!-- /br:models.chatVisible --> LLM models across 9 providers. It analyzes every request across 15 dimensions and routes it to the cheapest model that can handle it — in under 1ms, entirely locally.
+ClawRouter is an open-source local proxy that sits between your agent and <!-- br:models.chatVisible -->71<!-- /br:models.chatVisible --> LLM models across 9 providers. It analyzes every request across 15 dimensions and routes it to the cheapest model that can handle it — in under 1ms, entirely locally.
 
 ```
 Your Agent → ClawRouter (localhost:8402) → Best model for the job
@@ -74,7 +74,7 @@ Then point your agent at `http://localhost:8402/v1/` with any OpenAI-compatible
 
 - **No API keys to manage** — wallet-based cryptographic auth
 - **No subscriptions** — pay per request in USDC (Base or Solana)
-- **No vendor lock-in** — <!-- br:models.chatVisible -->70<!-- /br:models.chatVisible --> models, switch anytime
+- **No vendor lock-in** — <!-- br:models.chatVisible -->71<!-- /br:models.chatVisible --> models, switch anytime
 - **You control your wallet** — non-custodial, funds never held by a third party
 - **Budget caps** — set a max spend per session, ClawRouter gracefully downgrades when budget runs low
 - **Token compression** — 7-layer pipeline reduces token costs by 7–40% before they hit any provider

package/docs/clawrouter-cuts-llm-api-costs-500x.md

@@ -135,7 +135,7 @@ Less context repeated = fewer tokens = lower cost. Agents that need to recall ea
 
 ### 7. x402 Micropayments — Wallet-Based Budget Control
 
-<p align="center"><img src="assets/blockrun-clawrouter-x402-usdc-micropayment-wallet-budget-control.png" alt="Budget limits enforced by physical construction — wallet loaded via Base/Solana, pay per call across <!-- br:models.chatVisible -->70<!-- /br:models.chatVisible --> models, balance hits zero and the valve shuts cleanly" width="720"></p>
+<p align="center"><img src="assets/blockrun-clawrouter-x402-usdc-micropayment-wallet-budget-control.png" alt="Budget limits enforced by physical construction — wallet loaded via Base/Solana, pay per call across <!-- br:models.chatVisible -->71<!-- /br:models.chatVisible --> models, balance hits zero and the valve shuts cleanly" width="720"></p>
 
 ClawRouter pays for inference via [x402](https://x402.org/) USDC micropayments (Base or Solana). You load a wallet. Each inference call costs exactly what it costs. When the wallet runs low, requests stop cleanly.