This Q5'd model is now redundant - a fix was committed to the PR and it loads the Q6 models unconverted. A conversion of **mudler's Qwen3.6-35B-A3B-APEX-MTP** ("I-Quality", 23.5 GB) that runs on the **Strata** inference engine with PR #1402 (`qwen35moe` support). The model, its weights lineage and its in-file MTP draft layer are mudler's and Qwen's — this repo only changes the quantization of 27 tensors so the routed experts fit Strata's GPU kernels. **It is also a perfectly standard GGUF: llama.cpp and friends run it unchanged.** The APEX "I-Quality" ladder mixes routed-expert quantization per layer: 21 layers IQ4_XS, 10 layers Q5_K, **9 layers Q6_K**. Strata's engine (PR #1402) has GPU expert kernels for Q4_K / Q5_K / Q5_1 / Q8_0 / IQ2-4 families — but **not Q6_K experts** — so the original file is refused at load ("layer N's experts are Q6_K/Q6_K ... no GPU kernels for").

翻訳待ち・原文を表示

分類準備中

入力未確認→ 出力未確認

料金未確認マルチモーダル・チャネル料金は個別に確認

詳細を見る ↗

比較に追加出典・文書 ↗

Codestral-22B-v0.1

Mistral AI

> [!TIP] > PRs to correct the `transformers` tokenizer so that it gives 1-to-1 the same results as the `mistral_common` reference implementation are very welcome! Codestral-22B-v0.1 is trained on a diverse dataset of 80+ programming languages, including the most popular ones, such as Python, Java, C, C++, JavaScript, and Bash (more details in the Blogpost). The model can be queried: - As instruct, for instance to answer any questions about a code snippet (write documentation, explain, factorize) or to generate code following specific indications - As Fill in the Middle (FIM), to predict the middle tokens between a prefix and a suffix (very useful for software development add-ons like in VS Code) After installing `mistral_inference`, a `mistral-chat` CLI command should be available in your environment.

翻訳待ち・原文を表示

分類準備中

入力未確認→ 出力未確認

料金未確認マルチモーダル・チャネル料金は個別に確認

詳細を見る ↗

比較に追加出典・文書 ↗

Codestral 2508

Mistral AI

Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation. Blog Post

翻訳待ち・原文を表示

コード生成

入力テキスト · ファイル→ 出力テキスト

$0.3 / 100万入力トークンOpenRouter · mistralai/codestral-2508

詳細を見る ↗

比較に追加出典・文書 ↗

Devstral 2

Mistral AI

基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。

基本情報を収録・詳細は順次追加

テキスト生成

入力未確認→ 出力未確認

料金未確認マルチモーダル・チャネル料金は個別に確認

詳細を見る ↗

比較に追加出典・文書 ↗

Devstral 2 2512

Mistral AI

Devstral 2 is a state-of-the-art open-source model by Mistral AI specializing in agentic coding. It is a 123B-parameter dense transformer model supporting a 256K context window. Devstral 2 supports exploring...

翻訳待ち・原文を表示

テキスト生成オープンウェイト

入力テキスト · ファイル→ 出力テキスト

$0.4 / 100万入力トークンOpenRouter · mistralai/devstral-2512

詳細を見る ↗

比較に追加出典・文書 ↗