free-coding-models 0.5.32 → 0.5.34

package/LICENSE

@@ -1,6 +1,6 @@
-MIT License with NVIDIA Disclaimer
+MIT License with Enhanced Disclaimers
 
-Copyright (c) 2025 vava
+Copyright (c) 2025-2026 vava-nessa
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
@@ -22,28 +22,50 @@ SOFTWARE.
 
 ---
 
-NVIDIA DISCLAIMER
+GLOBAL DISCLAIMERS, PROVIDER GUARDRAILS & LIMITATION OF LIABILITY
+(DISCORDANCE, GARDE-FOUS PROVIDERS ET LIMITATION DE RESPONSABILITÉ)
 
-This project is NOT affiliated with, endorsed by, or sponsored by NVIDIA 
-Corporation.
+1. TRADEMARK, NOMINATIVE USE & NO AFFILIATION
+This software is an independent, community-developed open-source tool. It is NOT affiliated with, endorsed by, sponsored by, authorized by, or in any way officially connected with any of the third-party service providers referenced herein, including but not limited to: NVIDIA Corporation, Groq Inc., Cerebras Systems, SambaNova Systems Inc., Google LLC, GitHub Inc., Microsoft Corporation, Mistral AI, Alibaba Group, Cloudflare Inc., OVHcloud, Moonshot AI, MiniMax, or any other LLM provider.
+- All product names, logos, brands, and trademarks are the property of their respective owners. Their use in this software is strictly for descriptive and nominative purposes to indicate technical compatibility.
+- Nothing in this software or license shall be construed as creating a joint venture, partnership, agency, or employment relationship between the author and any third-party provider.
+- The author makes no claims of intellectual property over any provider's technology, branding, or services.
 
-"NVIDIA", "NIM", "NVIDIA NIM", and related trademarks are the property of 
-NVIDIA Corporation. Use of these trademarks is for descriptive purposes only 
-to indicate compatibility with NVIDIA's services.
+2. COMPLIANCE WITH PROVIDER TERMS OF SERVICE (ToS)
+Users of this software are solely and exclusively responsible for complying with the Terms of Service, Acceptable Use Policies, Developer Agreements, and Rate Limits of each third-party provider they interface with through this tool.
+- The author does not guarantee that using this tool complies with the terms of any provider.
+- The author shall not be held liable if the use of this software results in the suspension, restriction, or termination of the user's account by a provider, or in any other legal action taken by a provider against the user.
 
-This software is an independent, community-developed tool that interfaces 
-with NVIDIA's publicly documented APIs. The author makes no claims about 
-the availability, reliability, or performance of NVIDIA's services.
+3. INDEMNIFICATION BY THE USER (Indemnisation)
+By using this software, the user agrees to indemnify, defend, and hold harmless the author from and against any and all claims, liabilities, damages, losses, costs, expenses, or fees (including reasonable legal/attorneys' fees) arising out of or relating to:
+- The user's violation of any third-party provider's Terms of Service or policies.
+- Any dispute, litigation, or claim brought against the author by a third-party provider due to the user's deployment, misuse, or configuration of this software.
+- The user's configuration, storage, or transmission of API keys and credentials.
 
-Users of this software are solely responsible for:
-- Complying with NVIDIA's Terms of Service
-- Managing their own API keys securely
-- Any costs or usage associated with their NVIDIA account
+4. API EVOLUTION, DEPRECATION AND SERVICE INTERRUPTION
+The software interfaces with external, third-party APIs that are beyond the author's control.
+- Providers may modify, deprecate, rate-limit, monetize, or shut down their endpoints, models, or APIs at any time without notice.
+- The author is under no obligation to update, fix, or maintain the software to ensure continued compatibility with any provider's services.
+- The author shall not be liable for any service interruptions, data loss, or system failures caused by changes made by third-party providers.
 
-The author shall not be held liable for any issues arising from the use of 
-NVIDIA's services, including but not limited to service interruptions, 
-API changes, billing disputes, or account-related issues.
+5. API KEYS & CREDENTIALS SECURITY
+This software operates entirely locally on the user's machine.
+- API keys, tokens, and credentials (e.g., NVIDIA_API_KEY, GROQ_API_KEY, GITHUB_TOKEN, etc.) are stored locally on the user's device.
+- The author does not collect, store, transmit, or have access to the user's API keys or configuration files.
+- The user is solely responsible for securing their credentials. The author shall not be held liable for any credential leaks, unauthorized access, or security breaches resulting from the user's environment.
 
-For official NVIDIA NIM documentation and terms, visit:
-https://build.nvidia.com
-https://www.nvidia.com
+6. BILLING, COSTS, AND USAGE LIMITS
+Users are solely responsible for managing their accounts with third-party providers. The author shall not be liable for:
+- Any monetary charges, subscription fees, billing spikes, or financial damages incurred through the use of this software.
+- Rate limits, quota exhaustion, suspension, or termination of the user's accounts by third-party providers.
+
+7. AI GENERATED OUTPUTS & COMPLIANCE
+The software acts solely as an interface to connect to AI models. The author does not control, filter, or review the outputs generated by these models.
+- The user assumes all risks associated with the generated output (e.g., copyright infringement, license violations, software bugs, security vulnerabilities in generated code).
+- The author makes no warranties regarding the accuracy, safety, or legality of the code or content generated by the models.
+
+8. JURISDICTIONAL LIMITATION OF LIABILITY (French & EU Law / Droit Français et Européen)
+To the maximum extent permitted by applicable law, including but not limited to French Civil Code (Code civil) and EU consumer protection regulations:
+- Since this software is provided free of charge (à titre gratuit), the author's liability is excluded for all indirect, incidental, or consequential damages, including loss of profit, loss of data, loss of business, or system downtime.
+- The author’s liability shall only be engaged in the event of gross negligence (faute lourde) or intentional misconduct (dol), and is otherwise limited to the minimum extent allowed by law.
+- In any case, the author's total liability for any claim arising out of or in connection with this software shall not exceed €10 (ten Euros).

package/README.md

@@ -8,7 +8,7 @@
   <strong>Find the fastest free coding model in seconds</strong><br>
   Track ~191 models across 20 trusted free or free-limited AI providers in real time<br><br>
   <strong>Install Free API endpoints to your favorite AI coding tools:</strong><br>
-  OpenCode CLI / Desktop / WebUI, OpenClaw, Crush, Goose, Aider, Kilo CLI, Qwen Code, OpenHands, Amp, Hermes, Continue, Cline, Xcode, Pi and more...<br><br>
+  OpenCode CLI / Desktop / WebUI, OpenClaw, Crush, Goose, Aider, Kilo CLI, Qwen Code, OpenHands, Amp, Hermes, Continue, Cline, Xcode, Pi, ZCode and more...<br><br>
   <strong>Use Kimi K2, DeepSeek V3, GPT-OSS, Qwen3, MiniMax M3, GLM, Llama 4, Gemma 4, Devstral and more — for free</strong>
 </p>
 
@@ -431,6 +431,7 @@ Routing behavior:
 | `--pi` | π Pi |
 | `--copilot` | 🤖 Copilot CLI |
 | `--forgecode` | 🔥 ForgeCode |
+| `--zcode` | 🧊 ZCode |
 
 Press **`Z`** in the TUI to cycle between tools without restarting.
 

package/changelog/v0.5.33.md

@@ -0,0 +1,4 @@
+# Changelog v0.5.33 - 2026-06-17
+
+### Added
+- **ZCode tool mode (🧊).** Press `Z` in the TUI (or pass `--zcode`) to switch to ZCode mode. ZCode is the desktop agentic IDE from z.ai (智谱) that supports any OpenAI-compatible provider. Since ZCode stores its config in-app (Settings → Model Settings / 模型设置), FCM does not write a config file — instead, on launch it prints a step-by-step setup guide and launches the desktop app via `open -a ZCode` on macOS. The recommended setup points ZCode at the local FCM Router (`http://localhost:19280/v1`, API key `fcm-local`) so users get free failover and the upcoming GLM/Codestral per-provider normalizer on top of any FCM-tracked model. ZCode appears in the "Compatible with" column and the CLI `--help` tool list. Like Xcode, no auto-install is offered — download ZCode from zcode.z.ai/download.

package/changelog/v0.5.34.md

@@ -0,0 +1,4 @@
+# Changelog v0.5.34 - 2026-06-17
+
+### Fixed
+- **Per-provider schema normalization for GLM and Mistral/Codestral.** The router now runs a small normalizer before forwarding chat completions to `zai` (GLM) and `mistral` / `codestral`. The well-known 400/422 offenders are fixed at the router layer: `parallel_tool_calls`, `n`, `top_k`, `logprobs`, `echo`, `user`, `metadata`, and `store` are stripped; `tool` role messages that lack a matching assistant `tool_calls` entry are dropped (ZCode / Claude Code / Cline tool-call-flow quirk); and `temperature` is clamped to the provider's accepted range (Mistral: [0, 1]). Other OpenAI-compatible providers (Groq, Cerebras, NVIDIA, …) pass through unchanged. New module: `src/core/schema-normalizer.js`. Wired into both `proxyJsonRequest` and `proxyStreamingRequest`. 16 new tests in `test/test.js`.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "free-coding-models",
-  "version": "0.5.32",
+  "version": "0.5.34",
   "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",
@@ -69,7 +69,8 @@
     "chalk": "^5.6.2",
     "kandown": "^0.8.0",
     "socket.io": "^4.8.3",
-    "socket.io-client": "^4.8.3"
+    "socket.io-client": "^4.8.3",
+    "@tanstack/react-virtual": "^3.14.2"
   },
   "packageManager": "pnpm@10.33.2",
   "engines": {

package/sources.js

@@ -48,7 +48,6 @@ export const nvidiaNim = [
   ['deepseek-ai/deepseek-v4-flash',                'DeepSeek V4 Flash',   'S+', '72.0%', '1M'],
   ['stepfun-ai/step-3.5-flash',                    'Step 3.5 Flash',      'S+', '74.4%', '256k'],
   ['stepfun-ai/step-3.7-flash',                    'Step 3.7 Flash',      'S+', '74.4%', '256k'],
-  ['qwen/qwen3-coder-480b-a35b-instruct',          'Qwen3 Coder 480B',    'S+', '70.6%', '256k'],
   ['nvidia/nemotron-3-ultra-550b-a55b', 'Nemotron 3 Ultra', 'S+', '-', '1M'],
   // ── S tier — SWE-bench Verified 60–70% ──
   ['qwen/qwen3-next-80b-a3b-instruct',             'Qwen3 80B Instruct',  'S',  '65.0%', '128k'],
@@ -58,6 +57,7 @@ export const nvidiaNim = [
   ['mistralai/mistral-medium-3.5-128b',             'Mistral Medium 3.5',  'S',  '66.0%', '128k'],
   ['mistralai/mistral-small-4-119b-2603',           'Mistral Small 4',     'S',  '60.0%', '256k'],
   ['qwen/qwen3.5-122b-a10b',                       'Qwen3.5 122B',        'S',  '64.0%', '128k'],
+  ['minimaxai/minimax-m3',                         'MiniMax M3',          'S',  '59.0%', '1M'],
   // ── A+ tier — SWE-bench Verified 50–60% ──
   ['mistralai/mistral-large-3-675b-instruct-2512', 'Mistral Large 675B',  'A+', '58.0%', '256k'],
   ['nvidia/nemotron-3-super-120b-a12b',             'Nemotron 3 Super',    'A+', '56.0%', '128k'],
@@ -75,12 +75,6 @@ export const nvidiaNim = [
   ['meta/llama-3.2-11b-vision-instruct',           'Llama 3.2 11B Vision','B',  '28.0%', '128k'],
   // ── C tier — SWE-bench Verified <20% or lightweight edge models ──
   ['microsoft/phi-4-mini-instruct',                'Phi 4 Mini',          'C',  '14.0%', '128k'],
-  // Removed (2026-05-26): minimax-m2 (deprecated), qwen3-next-80b-a3b-thinking (deprecated),
-  // llama-3.1-nemotron-ultra-253b-v1 (404), llama-3.3-nemotron-super-49b-v1.5 (404),
-  // llama-3.3-70b-instruct (404), mixtral-8x22b-instruct-v0.1 (404),
-  // granite-34b-code-instruct (deprecated), llama-4-scout-17b-16e-instruct (deprecated, not in nvidiaNim),
-  // llama-3.1-8b-instruct (404) → replaced by llama-3.2-11b-vision-instruct
-  // Removed (2026-06-10): z-ai/glm5 (replaced by z-ai/glm-5.1)
 ]
 
 // 📖 Groq source - https://console.groq.com
@@ -101,8 +95,6 @@ export const groq = [
 export const cerebras = [
   ['gpt-oss-120b',                         'GPT OSS 120B',       'S',  '60.0%', '131k'],
   ['zai-glm-4.7',                          'GLM 4.7',            'S+', '73.8%', '131k'],
-  // Removed (2026-05-26): qwen-3-235b-a22b-instruct-2507 (deprecated May 27, 2026)
-  // Removed (2026-05-26): llama3.1-8b (deprecated May 27, 2026)
 ]
 
 // 📖 SambaNova source - https://cloud.sambanova.ai
@@ -114,14 +106,12 @@ export const sambanova = [
   // ── S tier ──
   ['DeepSeek-V3.1',                        'DeepSeek V3.1',      'S',  '62.0%', '128k'],
   ['DeepSeek-V3.2',                        'DeepSeek V3.2',      'S+', '70.0%', '32k'],
-  ['Llama-4-Maverick-17B-128E-Instruct',   'Llama 4 Maverick',   'S',  '62.0%', '128k'],
   ['gpt-oss-120b',                         'GPT OSS 120B',       'S',  '60.0%', '128k'],
+  // ── A tier ──
+  ['gemma-4-31B-it',                       'Gemma 4 31B',        'A',  '45.0%', '128k'],
   // ── A- tier ──
   ['Meta-Llama-3.3-70B-Instruct',          'Llama 3.3 70B',      'A-', '39.5%', '128k'],
   // ── B+ tier ──
-  ['gemma-3-12b-it',                        'Gemma 3 12B IT',     'B+', '46.0%', '128k'],
-  // Removed (2026-05-26): MiniMax-M2.5 (deprecated, replaced by MiniMax-M2.7)
-  // Fix (2026-05-26): Llama-4-Maverick context 1M → 128k (SambaNova actual limit)
 ]
 
 // 📖 OpenRouter source - https://openrouter.ai
@@ -137,9 +127,7 @@ export const sambanova = [
 export const openrouter = [
   // ── S+ tier — live :free chat/coding models ──
   ['qwen/qwen3-coder:free',                     'Qwen3 Coder 480B',   'S+', '70.6%', '1M'],
-  // Removed (2026-05-31): minimax/minimax-m2.5:free (free tier removed from OpenRouter)
   ['z-ai/glm-4.5-air:free',                     'GLM 4.5 Air',        'S+', '72.0%', '131k'],
-  // Removed (2026-05-31): deepseek/deepseek-v4-flash:free (free tier removed from OpenRouter)
   ['poolside/laguna-m.1:free',                  'Poolside Laguna M.1', 'S+', '-',     '262k'],
   ['poolside/laguna-xs.2:free',                 'Poolside Laguna XS.2','S+', '-',     '131k'],
   ['moonshotai/kimi-k2.6:free',                 'Kimi K2.6',          'S+', '76.8%', '262k'],
@@ -164,20 +152,11 @@ export const openrouter = [
   ['meta-llama/llama-3.2-3b-instruct:free',     'Llama 3.2 3B',       'B',  '20.0%', '128k'],
   // ── B+ tier ──
   ['nvidia/nemotron-nano-9b-v2:free',           'Nemotron Nano 9B',   'B+', '18.0%', '128k'],
-  // Removed (2026-05-31): baidu/cobuddy:free (free tier removed from OpenRouter)
   // ── B tier ──
   ['openrouter/free',                           'OpenRouter Free',    'B',  '-',     '200k'],
   // ── C tier ──
   ['liquid/lfm-2.5-1.2b-instruct:free',         'LFM 2.5 1.2B',       'C',  '-',     '32k'],
   ['liquid/lfm-2.5-1.2b-thinking:free',         'LFM 2.5 Thinking',   'C',  '-',     '32k'],
-  // Removed (2026-05-26): tencent/hy3-preview:free (free tier removed)
-  // Removed (2026-05-26): inclusionai/ling-2.6-1t:free (free tier removed)
-  // Removed (2026-05-26): google/gemma-3n-e2b-it:free (free tier removed)
-  // Removed (2026-05-26): google/gemma-3-27b-it:free (free tier removed)
-  // Removed (2026-05-26): google/gemma-3-12b-it:free (free tier removed)
-  // Removed (2026-05-26): google/gemma-3n-e4b-it:free (free tier removed)
-  // Removed (2026-05-26): google/gemma-3-4b-it:free (free tier removed)
-  // Removed (2026-05-26): google/gemma-4-31b-a4b-it:free (never existed — data error)
 ]
 
 // 📖 GitHub Models source - https://models.github.ai
@@ -210,9 +189,6 @@ export const mistral = [
   ['devstral-2512',             'Devstral 2',          'S+', '72.2%', '256k'],
   ['magistral-medium-2509',     'Magistral Medium 1.2','A+', '52.0%', '128k'],
   ['mistral-small-2603',        'Mistral Small 4',     'A',  '48.0%', '256k'],
-  // Removed (2026-05-26): devstral-medium-latest (deprecated, replaced by devstral-2512)
-  // Removed (2026-05-26): devstral-small-latest (deprecated, replaced by devstral-2512)
-  // Removed (2026-05-26): magistral-small-latest (deprecated, replaced by mistral-small-2603)
 ]
 
 // 📖 Mistral Codestral source - https://codestral.mistral.ai
@@ -231,6 +207,7 @@ export const scaleway = [
   ['qwen3-235b-a22b-instruct-2507',            'Qwen3 235B',         'S+', '70.0%', '250k'],
   ['gpt-oss-120b',                             'GPT OSS 120B',       'S',  '60.0%', '128k'],
   ['qwen3-coder-30b-a3b-instruct',             'Qwen3 Coder 30B',    'A+', '55.0%', '128k'],
+  ['qwen3.6-35b-a3b',                          'Qwen3.6 35B MoE',    'A+', '-',     '256k'],
   ['holo2-30b-a3b',                            'Holo2 30B',          'A+', '52.0%', '22k'],
   ['llama-3.3-70b-instruct',                   'Llama 3.3 70B',      'A-', '39.5%', '100k'],
   ['mistral-small-3.2-24b-instruct-2506',      'Mistral Small 3.2',  'B+', '30.0%', '128k'],
@@ -244,13 +221,9 @@ export const googleai = [
   ['gemini-3.1-pro-preview',                    'Gemini 3.1 Pro Preview',       'S+', '78.0%', '1M'],
   ['gemini-3-flash-preview',                    'Gemini 3 Flash Preview',       'S',  '65.0%', '1M'],
   ['gemini-3.1-flash-lite',                     'Gemini 3.1 Flash Lite',        'A+', '55.0%', '1M'],
-  // ⚠️ DEPRECATED — shutdown Oct 16, 2026
   ['gemini-2.5-pro',                            'Gemini 2.5 Pro',               'S+', '63.2%', '1M'],
-  // ⚠️ DEPRECATED — shutdown Oct 16, 2026
   ['gemini-2.5-flash',                          'Gemini 2.5 Flash',             'A+', '50.0%', '1M'],
-  // ⚠️ DEPRECATED — shutdown Oct 16, 2026
   ['gemini-2.5-flash-lite',                     'Gemini 2.5 Flash Lite',        'A',  '42.0%', '1M'],
-  // Removed (2026-05-26): gemini-3.1-flash-lite-preview (endpoint shutdown May 25, 2026 → replaced by gemini-3.1-flash-lite)
 ]
 
 // 📖 ZAI source - https://open.z.ai
@@ -281,7 +254,6 @@ export const qwen = [
   ['qwen3.5-flash',                           'Qwen3.5 Flash',     'A+', '55.0%', '1M'],
   ['qwen3-coder-flash',                       'Qwen3 Coder Flash', 'A+', '55.0%', '1M'],
   ['qwen3-32b',                               'Qwen3 32B',          'A+', '50.0%', '256k'],
-  // Removed (2026-05-26): qwen2.5-coder-32b-instruct (legacy, replaced by qwen3-coder-plus/flash)
   // Fix (2026-05-26): qwen3-max ctx 1M→256k, qwen3-coder-plus ctx 256k→1M, qwen3-coder-flash ctx 256k→1M, qwen3-32b ctx 128k→256k
   // Fix (2026-05-26): qwen3-235b-a22b-instruct ID → qwen3-235b-a22b
 ]
@@ -293,18 +265,15 @@ export const qwen = [
 export const cloudflare = [
   // ── S+ tier ──
   ['@cf/moonshotai/kimi-k2.6',                'Kimi K2.6',         'S+', '76.8%', '262k'],
+  ['@cf/moonshotai/kimi-k2.7-code',           'Kimi K2.7 Code',    'S+', '-',     '262k'],
   // ── S tier ──
   ['@cf/zai-org/glm-4.7-flash',               'GLM-4.7-Flash',     'S',  '59.2%', '131k'],
+  ['@cf/openai/gpt-oss-120b',                 'GPT OSS 120B',      'S',  '60.0%', '128k'],
   // ── A+ tier ──
-  // Removed (2026-05-31): @cf/qwen/qwq-32b (upstream deprecation — Groq removed qwen-qwq-32b July 2025)
   ['@cf/nvidia/nemotron-3-120b-a12b',         'Nemotron 3 Super',  'A+', '56.0%', '32k'],
   // ── A tier ──
   ['@cf/meta/llama-4-scout-17b-16e-instruct', 'Llama 4 Scout',     'A',  '44.0%', '131k'],
-  ['@cf/meta/llama-3.2-90b-instruct',          'Llama 3.2 90B',     'A+', '45.0%', '128k'],
   ['@cf/mistralai/mistral-7b-instruct-v0.2',   'Mistral 7B v0.2',   'A',  '38.0%', '128k'],
-  ['@cf/google/gemma-2-9b-it',                 'Gemma 2 9B',        'A',  '35.0%', '128k'],
-  ['@cf/anthropic/claude-3-5-sonnet',         'Claude 3.5 Sonnet',  'S+', '62.4%', '200k'],
-  ['@cf/openai/gpt-4o-mini',                   'GPT-4o Mini',       'A+', '50.0%', '128k'],
   ['@cf/qwen/qwen3-30b-a3b-fp8',              'Qwen3 30B MoE',     'A',  '45.0%', '128k'],
   ['@cf/qwen/qwen2.5-coder-32b-instruct',     'Qwen2.5 Coder 32B', 'A',  '46.0%', '128k'],
   ['@cf/openai/gpt-oss-20b',                  'GPT OSS 20B',       'A',  '42.0%', '128k'],
@@ -315,8 +284,6 @@ export const cloudflare = [
 
   // ── B tier ──
   ['@cf/ibm/granite-4.0-h-micro',             'Granite 4.0 Micro', 'B+', '30.0%', '128k'],
-  // Removed (2026-05-26): @cf/google/gemma-4-31b-it (never existed on Cloudflare)
-  // Removed (2026-05-26): @cf/meta/llama-3.1-8b-instruct (deprecated May 30 + actual ctx = 8k, not 128k)
   // Fix (2026-05-26): @cf/moonshotai/kimi-k2.6 ctx 256k → 262k
 ]
 
@@ -326,6 +293,7 @@ export const cloudflare = [
 // 📖 Env var: OVH_AI_ENDPOINTS_ACCESS_TOKEN
 export const ovhcloud = [
   ['Qwen3.5-397B-A17B',                         'Qwen3.5 397B MoE',    'S',  '-',     '262k'],
+  ['Qwen3.6-27B',                               'Qwen3.6 27B',         'A',  '-',     '262k'],
   ['Qwen3-Coder-30B-A3B-Instruct',             'Qwen3 Coder 30B MoE',  'A+', '55.0%', '256k'],
   ['gpt-oss-120b',                              'GPT OSS 120B',         'S',  '60.0%', '131k'],
   ['gpt-oss-20b',                               'GPT OSS 20B',          'A',  '42.0%', '131k'],
@@ -335,7 +303,6 @@ export const ovhcloud = [
   ['Mistral-7B-Instruct-v0.3',                  'Mistral 7B Instruct',  'B',  '25.0%', '127k'],
   ['Mistral-Nemo-Instruct-2407',                'Mistral Nemo',         'B+', '30.0%', '118k'],
   ['Qwen3.5-9B',                                'Qwen3.5 9B',           'B+', '30.0%', '262k'],
-  // Removed (2026-05-26): Llama-3.1-8B-Instruct (delisted from OVHcloud catalog)
   // Fix (2026-05-26): Qwen3.5-9B ctx 128k→262k, Mistral-Small ctx 131k→128k, Mistral-Nemo ctx 128k→118k, Mistral-7B ctx 32k→127k
 ]
 
@@ -350,17 +317,8 @@ export const opencodeZen = [
   ['big-pickle',                       'Big Pickle',              'S+', '72.0%', '200k'],
   ['deepseek-v4-flash-free',           'DeepSeek V4 Flash Free',  'S+', '79.0%', '200k'],
   ['mimo-v2.5-free',                   'MiMo-V2.5 Free',          'S+', '-',     '200k'],
-  ['nemotron-3-super-free',            'Nemotron 3 Super Free',   'A+', '52.0%', '200k'],
-  ['minimax-m3-free',                  'MiniMax M3 Free',         'S+', '59.0%', '1M'],
-  // Removed (2026-05-28): qwen3.6-plus-free (free promotion ended — now requires OpenCode Go)
-  // Removed (2026-05-28): minimax-m2.5-free (7-day launch promo expired — now paid)
-  // Removed (2026-05-28): gpt-5-nano (was never free — incorrectly cataloged as free, requires Zen PAYG)
-  // Removed (2026-05-26): hy3-preview-free (deleted from Zen)
-  // Removed (2026-05-26): ling-2.6-flash-free (deleted from Zen)
-  // Removed (2026-05-26): trinity-mini-free (deleted from Zen)
-  // Removed (2026-05-26): trinity-large-preview-free (deleted from Zen)
-  // Fix (2026-05-26): nematron-3-super-free ctx 1M → 200k (Zen limits it)
-  // Added (2026-06-01): minimax-m3-free — MiniMax M3, 1M context, MSA attention (SWE-Bench Pro 59.0%)
+  ['nemotron-3-ultra-free',            'Nemotron 3 Ultra Free',   'A+', '-',     '200k'],
+  ['north-mini-code-free',             'North Mini Code Free',    'B+', '-',     '200k'],
 ]
 
 // 📖 Kilo source - https://api.kilo.ai/api/gateway
@@ -374,10 +332,12 @@ export const kilo = [
 // 📖 Free unauthenticated tier works with tight shared limits; optional free token at https://token.llm7.io
 // 📖 Pro-tagged models from /v1/models are intentionally excluded.
 export const llm7 = [
+  ['deepseek-v4-flash',                      'DeepSeek V4 Flash',   'S+', '-',     '128k'],
+  ['kimi-k2.6',                              'Kimi K2.6',           'S+', '-',     '256k'],
   ['qwen3-235b',                             'Qwen3 235B',          'S+', '70.0%', '240k'],
+  ['devstral-small-2:24b',                   'Devstral Small 2',    'A',  '-',     '128k'],
   ['mistral-small-3.2',                      'Mistral Small 3.2',   'B+', '34.0%', '128k'],
   ['codestral-latest',                       'Codestral Latest',    'B+', '34.0%', '32k'],
-  ['devstral-small-2:24b',                   'Devstral Small 2',    'A',  '-',     '128k'],
 ]
 
 // 📖 Routeway source - https://api.routeway.ai/v1/models