free-coding-models 0.5.84 → 0.5.86

package/README.md

@@ -192,6 +192,7 @@ A large catalog of coding models from 20+ active providers, ranked by practical
 | 21 | [SiliconFlow](https://cloud.siliconflow.cn/account/ak) | 3 | A → B+ | 3 models @ $0, 1000 RPM (phone SMS required) | `SILICONFLOW_API_KEY` |
 | 22 | [Requesty](https://app.requesty.ai/api-keys) | 12 | S+ → C | 200 req/day free, no card (4x OpenRouter) | `REQUESTY_API_KEY` |
 | 23 | [OrcaRouter](https://www.orcarouter.ai) | 3 | S+ → A+ | Free Hacker tier · 3 $-0 models, zero token markup | `ORCAROUTER_API_KEY` |
+| 24 | [Vercel AI Gateway](https://vercel.com/ai-gateway) | 4 | S+ → B+ | $5 gateway credits every 30 days, no card + explicit $0 models | `VERCEL_AI_GATEWAY_API_KEY` |
 
 > **Pollinations Pollen, c'est quoi ?** Pollinations ne facture pas en $ mais en **Pollen** (≈ $1). Pas besoin d'acheter : tu recois du Pollen **gratuit chaque jour** selon ton tier seed/flower/nectar via quests (star le repo `pollinations/pollinations` sur GitHub, aider sur un issue). Si ton solde affiche `0.0000` sur `enter.pollinations.ai`, tout `MISS` (prompt non cache) retourne `402 PAYMENT_REQUIRED` ("Insufficient balance ~0.0001 pollen") et le ping affichera `402`. Un `HIT` Cloudflare cache peut encore passer en `200` meme avec 0 Pollen mais ce n'est pas fiable. Verifie ton solde sur `enter.pollinations.ai` -> API Keys, complete une quest et attends le grant quotidien a minuit UTC. **Note : integration Pollinations encore experimentale et non confirmee comme stable, le systeme Pollen peut changer.**
 > **SiliconFlow & Requesty** : integrations verifiees le 2026-08-24 via `/v1/models` live et docs officielles, free tiers recurrents confirmes sans Pollen.
@@ -985,6 +986,7 @@ Curated resources kept **outside** the active CLI catalog — IDE extensions, co
 
 | Provider | Free limits | Notable models |
 |----------|-------------|----------------|
+| [Vercel AI Gateway](https://vercel.com/ai-gateway) | $5 credits every 30 days (no card) + explicit $0 models | MiniMax M3 (1M ctx), Laguna S 2.1, MiniMax M2.7, Ling 3.0 Flash Fin |
 | [OrcaRouter](https://www.orcarouter.ai) | Free Hacker tier, zero token markup, 3 API keys | DeepSeek V4 Flash (Free), Tencent Hy3 (Free), Qwen3.8 27B (Free) |
 | [OpenRouter](https://openrouter.ai/keys) | 50 req/day, 1K/day with $10 purchase | Qwen3-Coder, Tencent HY3, Laguna, Gemma 4 |
 | [Google AI Studio](https://aistudio.google.com/apikey) | Varies by Gemini model and region | Gemini 3.8 Flash, Gemini 2.5 Pro |

package/changelog/v0.5.85.md

@@ -0,0 +1,21 @@
+# Changelog v0.5.85 - 2026-09-05
+
+### Added
+
+- New provider: **Vercel AI Gateway** (24th provider in the catalog). This is Vercel's official OpenAI-compatible gateway: one endpoint at `https://ai-gateway.vercel.sh/v1/chat/completions`, list prices with zero markup, and it works from your local machine with nothing deployed to Vercel.
+  - **Free tier**: every Vercel account gets **$5 of gateway credits every 30 days**, no credit card required. On top of that, the gateway exposes a handful of genuinely $0 models (input AND output priced at $0), and those are the ones we list.
+  - **4 models added**, all verified live against `GET /v1/models` (373 models scanned, 2026-09-05):
+    - `minimax/minimax-m3-free` - MiniMax M3 (Free), S+ 78.4%, 1M context, vision + tool use
+    - `minimax/minimax-m2.7-free` - MiniMax M2.7 (Free), S+ 78.0%, 196k context, software-engineering focused
+    - `poolside/laguna-s-2.1-free` - Laguna S 2.1 (Free), S+ tier, 256k context, dedicated agentic coding model
+    - `inclusionai/ling-3.0-flash-fin-free` - Ling 3.0 Flash Fin (Free), B+ tier, 256k context, finance-tuned 124B MoE that retains coding and math
+  - **Quota marked `limited`** on purpose: the monthly credit only covers a subset of the 373-model catalog, and per Vercel's FAQ purchasing credits once permanently moves the account to the paid tier, which kills the recurring free credit. The Settings panel explains this.
+  - **Full cross-surface wiring**: TUI Settings panel entry with its own theme color, key test endpoint (`/v1/models`), web dashboard catalog, OpenCode + Kilo provider config generation, and monthly reset cadence for quota display. Env var: `VERCEL_AI_GATEWAY_API_KEY` (also accepts `AI_GATEWAY_API_KEY`).
+  - Get a key: Vercel dashboard, AI Gateway → API Keys.
+- New unit test covering the `vercel-gateway` source entry structure (model IDs must be explicit `-free` $0 models). Suite now at 880 tests.
+
+### Notes
+
+- MiniMax M3/M2.7 scores mirror the measured entries already in the catalog (ollama-cloud, benchmarked 2026-07-27); the underlying models are identical.
+- Laguna S 2.1 has no published SWE-bench score; its S+ tier follows the family precedent (Laguna XS 2.1 ships S+ 70.9% via NVIDIA). models.dev flags the model as deprecated as of 2026-09-05 (the Zen free promo ended), but Vercel still serves the $0 variant live today, so it stays listed with a re-verify note for the next audit.
+- Catalog totals: 24 providers, 229 models.

package/changelog/v0.5.86.md

@@ -0,0 +1,10 @@
+# Changelog v0.5.86 - 2026-09-05
+
+### Fixed
+- 🖥️ Provider column in the TUI table widened from 11 to 15 characters, with automatic truncation at 14 characters (13 + ellipsis) for anything longer. Provider names like "Cloudflare AI" (13 chars) now render fully instead of pushing every following cell out of alignment. Any future provider name longer than 14 chars is truncated cleanly with an ellipsis, so a long label can never break the table layout again. The compact-mode truncation (4 chars + ellipsis on narrow terminals) is unchanged.
+- ⚡ Command Palette no longer paints a giant dark rectangle over the table. Each palette row used to be emitted from column 1 with a leading margin of unstyled spaces plus a trailing "erase to end of line": both inherited the still-active panel background, so the panel tint filled the entire screen width for every palette row, masking the frozen table on both sides of the modal. Rows are now positioned directly at the panel's left column and paint only the panel's own cells; the frozen table stays visible beside the palette.
+- ✨ Command Palette no longer flickers on large terminals. While the palette is open the table behind it is frozen, so repainting the full table every frame was pure waste: on a 280-column terminal the per-frame write was ~300 KB, got split into chunks, and the terminal showed the bare table for a moment before the palette painted over it, 30 times per second. The frozen table is now written once when the palette opens; after that only the palette itself is repainted (per-frame output divided by ~50).
+- 🧹 Palette height now stays constant while typing a search. Nothing repaints the cells below the panel while it is open (the table behind is frozen), so a shrinking result list used to leave stale painted rows hanging under the panel's bottom edge. Short result lists are padded to the full body height instead.
+
+### Tests
+- 2 new regression tests: the palette must never emit margin filler or edge-to-edge erase codes, and the panel must keep a constant height when the result list shrinks. 883 tests passing across 11 suites.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "free-coding-models",
-  "version": "0.5.84",
+  "version": "0.5.86",
   "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.",
   "keywords": [
     "nvidia",

package/sources.js

@@ -29,7 +29,7 @@
  *   📖 Secondary: https://swe-rebench.com (independent evals, scores are lower)
  *   📖 Leaderboard tracker: https://www.marc0.dev/en/leaderboard
  *
- *   @exports nvidiaNim, groq, cerebras, sambanova, openrouter, githubModels, mistral, codestral, scaleway, googleai, zai, qwen, cloudflare, ovhcloud, opencodeZen, kilo, llm7, routeway, novita, ollamaCloud, pollinations, siliconflow, requesty, orcarouter — model arrays per active provider
+ *   @exports nvidiaNim, groq, cerebras, sambanova, openrouter, githubModels, mistral, codestral, scaleway, googleai, zai, qwen, cloudflare, ovhcloud, opencodeZen, kilo, llm7, routeway, novita, ollamaCloud, pollinations, siliconflow, requesty, orcarouter, vercelGateway — model arrays per active provider
  *   @exports sources — map of active free/free-limited providers, each with { name, url, models }
 
  *   @exports MODELS — flat array of [modelId, label, tier, sweScore, ctx, providerKey]
@@ -564,6 +564,24 @@ export const orcarouter = [
   ['qwen/qwen3.8-27b-free', 'Qwen3.8 27B (Free)', 'A+', '-', '64k'],
 ]
 
+// 📖 Vercel AI Gateway source - https://vercel.com/docs/ai-gateway
+// 📖 OpenAI-compatible gateway: https://ai-gateway.vercel.sh/v1/chat/completions
+// 📖 Official Vercel gateway at list prices (zero markup). Every account gets $5 of
+// 📖 gateway credits every 30 days (no card needed), and the catalog also exposes a
+// 📖 handful of genuinely $0 models (input AND output priced 0). Verified live
+// 📖 2026-09-05 via GET /v1/models (373 models, 5 with $0/$0 pricing).
+// 📖 Caveats: the monthly credit only covers a subset of the catalog, and buying
+// 📖 credits once permanently moves the account to the paid tier (official FAQ),
+// 📖 which is why this provider is quotaCode 'limited'.
+export const vercelGateway = [
+  // ── S+ tier — SWE-bench Verified ≥70% ──
+  ['minimax/minimax-m3-free', 'MiniMax M3 (Free)', 'S+', '78.4%', '1M'], // score mirrors minimax-m3 (ollama-cloud, measured 2026-07-27)
+  ['minimax/minimax-m2.7-free', 'MiniMax M2.7 (Free)', 'S+', '78.0%', '196k'], // score mirrors minimax-m2.7 (ollama-cloud, measured 2026-07-27)
+  ['poolside/laguna-s-2.1-free', 'Laguna S 2.1 (Free)', 'S+', '-', '256k'], // tier follows family precedent: laguna-xs-2.1 ships S+ 70.9% via NVIDIA. Caution: models.dev flags laguna-s-2.1 deprecated (2026-09-05, Zen promo ended) but Vercel still serves the $0 variant live - re-verify at next audit
+  // ── B+ tier — vertical-tuned lightweight (coding secondary) ──
+  ['inclusionai/ling-3.0-flash-fin-free', 'Ling 3.0 Flash Fin (Free)', 'B+', '-', '256k'], // 124B MoE (5.1B active), finance-tuned, retains coding + math
+]
+
 // 📖 Ollama Cloud source - https://ollama.com/pricing and https://ollama.com/search?c=cloud
 // 📖 Free plan includes cloud model access with session/weekly limits. This list keeps coding-relevant cloud models only.
 // 📖 Catalog verified 2026-07-18 against official Ollama cloud model search page.
@@ -773,6 +791,13 @@ export const sources = {
     quotaCode: 'free',
     models: orcarouter,
   },
+  'vercel-gateway': {
+    name: 'Vercel AI Gateway',
+    url: 'https://ai-gateway.vercel.sh/v1/chat/completions',
+    quota: 'Free · $5 credits/30 days + $0 models · no card',
+    quotaCode: 'limited',
+    models: vercelGateway,
+  },
   'ollama-cloud': {
     name: 'Ollama Cloud',
     url: 'https://ollama.com/v1/chat/completions',

package/src/core/config.js

@@ -187,6 +187,7 @@ const ENV_VARS = {
   siliconflow: 'SILICONFLOW_API_KEY',
   requesty:   'REQUESTY_API_KEY',
   orcarouter: 'ORCAROUTER_API_KEY',
+  'vercel-gateway': ['VERCEL_AI_GATEWAY_API_KEY', 'AI_GATEWAY_API_KEY'],
 }
 
 // 📖 Smart Router defaults are intentionally conservative: balanced probing,

package/src/core/kilo.js

@@ -117,6 +117,13 @@ export async function startKilo(model, fcmConfig) {
         options: { baseURL: 'https://api.orcarouter.ai/v1', apiKey: '{env:ORCAROUTER_API_KEY}' },
         models: {}
       }
+    } else if (providerKey === 'vercel-gateway') {
+      config.provider['vercel-gateway'] = {
+        npm: '@ai-sdk/openai-compatible',
+        name: 'Vercel AI Gateway',
+        options: { baseURL: 'https://ai-gateway.vercel.sh/v1', apiKey: '{env:VERCEL_AI_GATEWAY_API_KEY}' },
+        models: {}
+      }
     } else if (providerKey === 'huggingface') {
       config.provider.huggingface = {
         npm: '@ai-sdk/openai-compatible',

package/src/core/models-dev-index.js

@@ -45,6 +45,7 @@ export const PROVIDER_ALIASES = {
   nvidia:     'nvidia',
   openrouter: 'openrouter',
   orcarouter: 'orcarouter',
+  'vercel-gateway': 'vercel',
   groq:       'groq',
   cerebras:   'cerebras',
   github:     'github-models',

package/src/core/opencode.js

@@ -432,6 +432,13 @@ export async function startOpenCode(model, fcmConfig) {
         options: { baseURL: 'https://api.orcarouter.ai/v1', apiKey: '{env:ORCAROUTER_API_KEY}' },
         models: {}
       }
+    } else if (providerKey === 'vercel-gateway') {
+      config.provider['vercel-gateway'] = {
+        npm: '@ai-sdk/openai-compatible',
+        name: 'Vercel AI Gateway',
+        options: { baseURL: 'https://ai-gateway.vercel.sh/v1', apiKey: '{env:VERCEL_AI_GATEWAY_API_KEY}' },
+        models: {}
+      }
     } else if (providerKey === 'huggingface') {
       config.provider.huggingface = {
         npm: '@ai-sdk/openai-compatible',
@@ -807,6 +814,13 @@ export async function startOpenCodeDesktop(model, fcmConfig) {
         options: { baseURL: 'https://api.orcarouter.ai/v1', apiKey: '{env:ORCAROUTER_API_KEY}' },
         models: {}
       }
+    } else if (providerKey === 'vercel-gateway') {
+      config.provider['vercel-gateway'] = {
+        npm: '@ai-sdk/openai-compatible',
+        name: 'Vercel AI Gateway',
+        options: { baseURL: 'https://ai-gateway.vercel.sh/v1', apiKey: '{env:VERCEL_AI_GATEWAY_API_KEY}' },
+        models: {}
+      }
     } else if (providerKey === 'huggingface') {
       config.provider.huggingface = {
         npm: '@ai-sdk/openai-compatible',

package/src/core/provider-key-tester.js

@@ -69,6 +69,7 @@ export const PROVIDER_AUTH_ENDPOINTS = {
   pollinations: null,
   requesty:     null,
   orcarouter:   { url: 'https://api.orcarouter.ai/v1/models',   method: 'GET' },
+  'vercel-gateway': { url: 'https://ai-gateway.vercel.sh/v1/models', method: 'GET' },
   together:     { url: 'https://api.together.xyz/v1/models',        method: 'GET' },
   perplexity:   { url: 'https://api.perplexity.ai/v1/models',       method: 'GET' },
   chutes:       { url: 'https://chutes.ai/v1/models',               method: 'GET' },

package/src/core/provider-metadata.js

@@ -66,6 +66,7 @@ export const ENV_VAR_NAMES = {
   siliconflow: 'SILICONFLOW_API_KEY',
   requesty:   'REQUESTY_API_KEY',
   orcarouter: 'ORCAROUTER_API_KEY',
+  'vercel-gateway': 'VERCEL_AI_GATEWAY_API_KEY',
 }
 
 // 📖 OPENCODE_MODEL_MAP: sparse table of model IDs that differ between sources.js and OpenCode's
@@ -339,6 +340,14 @@ export const PROVIDER_METADATA = {
     rateLimits: 'Free Hacker tier · zero token markup · 3 API keys',
     detailedLimits: 'Zero-markup AI gateway: token prices are passed through at provider rates.\nFree Hacker plan: 3 API keys, adaptive routing + automatic failover + guardrails included.\nOnly the explicit $-0 models are listed in this catalog; the orcarouter/fusion family is pay-as-you-go.',
   },
+  'vercel-gateway': {
+    label: 'Vercel AI Gateway',
+    color: chalk.rgb(140, 170, 255),
+    signupUrl: 'https://vercel.com/ai-gateway',
+    signupHint: 'AI Gateway → API Keys ($5 credits/30 days, no card)',
+    rateLimits: 'Free tier · $5 credits/30 days + $0 models',
+    detailedLimits: 'Official Vercel gateway at list prices, zero markup.\nEvery account gets $5 of credits every 30 days, no card required.\nThe monthly credit covers a subset of the catalog; the explicit $0 models stay free.\nCaution: purchasing credits once permanently moves the account to the paid tier and the monthly free credit stops.',
+  },
   'ollama-cloud': {
     label: 'Ollama Cloud',
     color: chalk.rgb(230, 230, 230),

package/src/core/quota-capabilities.js

@@ -65,6 +65,8 @@ export const PROVIDER_CAPABILITIES = {
   siliconflow: { telemetryType: 'unknown', supportsEndpoint: false, usageDisplay: 'ok', resetCadence: 'daily' },
   requesty: { telemetryType: 'unknown', supportsEndpoint: false, usageDisplay: 'ok', resetCadence: 'daily' },
   orcarouter: { telemetryType: 'unknown', supportsEndpoint: false, usageDisplay: 'ok', resetCadence: 'unknown' },
+  // 📖 Vercel AI Gateway exposes GET /v1/credits but no fetcher is wired yet; treat as monthly reset.
+  'vercel-gateway': { telemetryType: 'unknown', supportsEndpoint: false, usageDisplay: 'ok', resetCadence: 'monthly' },
 }
 
 /** Fallback for unrecognized providers */

package/src/tui/app.js

@@ -1075,8 +1075,15 @@ export async function runApp(cliArgs, config, startupOptions = {}) {
         // 📖 Freeze the full table (including countdown and spinner glyphs) while
         // 📖 the command palette is open so the background remains perfectly static.
         state.commandPaletteFrozenTable = renderTable(tableOpts)
+        tableContent = state.commandPaletteFrozenTable
+      } else {
+        // 📖 The frozen table is already on screen and cannot change while the
+        // 📖 palette is open, so repaint the palette alone. Repainting the full
+        // table every frame made large terminals flicker: the write gets split
+        // into chunks, so the terminal showed the bare table for a moment
+        // before the palette overlay painted over it, 30 times per second.
+        tableContent = null
       }
-      tableContent = state.commandPaletteFrozenTable
     } else {
       state.commandPaletteFrozenTable = null
       tableContent = renderTable(tableOpts)
@@ -1101,7 +1108,7 @@ export async function runApp(cliArgs, config, startupOptions = {}) {
       : state.incompatibleFallbackOpen
         ? overlays.renderIncompatibleFallback()
       : state.commandPaletteOpen
-        ? tableContent + overlays.renderCommandPalette()
+        ? (tableContent || '') + overlays.renderCommandPalette()
       : state.recommendOpen
         ? overlays.renderRecommend()
         : state.helpVisible

package/src/tui/overlays.js

@@ -856,6 +856,11 @@ export function createOverlayRenderers(state, deps) {
     }
 
     const targetLine = cursorLineByRow[state.commandPaletteCursor] ?? 0
+    // 📖 Pad short result lists up to bodyRows so the panel height stays constant
+    // 📖 while typing: nothing repaints the cells below a shrunken panel (the
+    // 📖 table behind is frozen while the palette is open), so a shrinking panel
+    // 📖 would leave stale painted rows hanging under its bottom edge.
+    while (panelLines.length < bodyRows) panelLines.push('')
     state.commandPaletteScrollOffset = keepOverlayTargetVisible(
       state.commandPaletteScrollOffset,
       targetLine,
@@ -922,11 +927,16 @@ export function createOverlayRenderers(state, deps) {
       return sty.overlayBgCommandPalette(padded)
     })
 
-    // 📖 Emit each row from column 1 and clear to end of line so stale table cells
-    // 📖 in the margins beside the panel can never bleed through (the artifacts
-    // 📖 reported in issue #169). \x1b[K never wraps: panel right edge <= cols.
+    // 📖 Emit each row directly at the panel's left column: no margin filler and
+    // 📖 no \x1b[K. The old form (col 1 + margin spaces + \x1b[K) painted the
+    // 📖 panel tint across the full screen width because the margin spaces and
+    // 📖 the erase-to-EOL both inherit the tint background still active from the
+    // 📖 previous row. On wide terminals that grew a giant dark rectangle over
+    // 📖 the frozen table on both sides of the panel, repainted every frame.
+    // 📖 Rows are width-budgeted and the panel is screen-clamped, so a row can
+    // 📖 never overflow past the terminal edge (the real issue #169 artifact).
     return tintedLines
-      .map((line, idx) => `\x1b[${top + idx};1H${' '.repeat(left - 1)}${line}\x1b[K`)
+      .map((line, idx) => `\x1b[${top + idx};${left}H${line}`)
       .join('')
   }
 

package/src/tui/render-table.js

@@ -314,7 +314,13 @@ export function renderTable({
   const W_RANK = 6
   const W_TIER = 5
   const W_CTX = 4
-  const W_SOURCE = 11
+  const W_SOURCE = 15
+  // 📖 Provider names are display-capped at MAX_PROVIDER_NAME_LEN chars so an
+  // 📖 over-long label can never push the following cells and desalign the
+  // 📖 whole row (Cloudflare AI at 13 chars overflowed the old 11-wide column).
+  // 📖 Anything longer is cut to 13 chars + '…' = 14 visible chars. Keep new
+  // 📖 provider names in sources.js at MAX_PROVIDER_NAME_LEN chars or less.
+  const MAX_PROVIDER_NAME_LEN = 14
   const W_MODEL = 26
   const W_SWE = 5
   const W_STATUS = 17
@@ -719,11 +725,15 @@ export function renderTable({
     const tier = tierFn(r.tier.padEnd(W_TIER))
     // 📖 Keep terminal view provider-specific so each row is monitorable per provider
     // 📖 In compact mode, truncate provider name to 4 chars + '…'
+    // 📖 In normal mode, cap names at MAX_PROVIDER_NAME_LEN chars (13 + '…')
+    // 📖 so an over-long label never shifts every cell to its right.
     const providerNameRaw = sources[r.providerKey]?.name ?? r.providerKey ?? 'NIM'
     const providerName = normalizeOriginLabel(providerNameRaw, r.providerKey)
     const providerDisplay = isCompact && providerName.length > 5
       ? providerName.slice(0, 4) + '…'
-      : providerName
+      : providerName.length > MAX_PROVIDER_NAME_LEN
+        ? providerName.slice(0, MAX_PROVIDER_NAME_LEN - 1).trimEnd() + '…'
+        : providerName
     const source = themeColors.provider(r.providerKey, providerDisplay.padEnd(wSource))
     // 📖 Prefix: ⭐ favorite > 🎯 recommended > 🆕 new — only one emoji, never shifts the line
     const modelIsNew = isNewModel(r.addedDate)

package/src/tui/theme.js

@@ -195,6 +195,7 @@ const PROVIDER_PALETTES = {
     pollinations: [255, 105, 180],
     requesty: [100, 149, 255],
     orcarouter: [255, 138, 64],
+    'vercel-gateway': [170, 190, 230],
     'ollama-cloud': [230, 230, 230],
   },
   light: {
@@ -228,6 +229,7 @@ const PROVIDER_PALETTES = {
     pollinations: [170, 45, 110],
     requesty: [0, 72, 170],
     orcarouter: [170, 74, 0],
+    'vercel-gateway': [60, 80, 130],
     'ollama-cloud': [88, 88, 88],
   },
 }