In short
How they differ.
Worked out from the figures below. Prices and specs are the makers' own, read Oct 20, 2026.
- GPT-6.1 Sol reads more at once: a context window of 1M tokens against 262K.
- Qwen3.8-2.4T-A95B can be downloaded and run on your own machine; GPT-6.1 Sol can't.
- No independent board we follow ranks both models yet.
Side by side
The specs.
| GPT-6.1 Sol | Qwen3.8-2.4T-A95B | |
|---|---|---|
| Made by | OpenAI · US | Alibaba (Qwen) · China |
| Released | Sep 29, 2026 | Aug 8, 2026 |
| Price per 1M tokens | $2 in / $10 out | Free to download; hosts set prices |
| Context window | 1M | 262K |
| Takes in | text, image | text |
| Download and run it yourself | No | Yes (qwen3.8-max (license tag shown on Hugging Face)) |
| Size | Not published | 2.4T parameters, 95B active |
| Maker says | “Near-Astra performance for complex work at a lower cost.” | “For the first time, Qwen3.8 brings a Qwen-Max-class model to open release.” |
| Sources | Prices, Oct 20, 2026 · Description | Prices, Oct 11, 2026 · Description |
Independent tests
Where they place.
Position on each board we follow (1 is best). Epoch AI's and Vectara's scores are shown; the other boards are linked for their own figures.
| Board | GPT-6.1 Sol | Qwen3.8-2.4T-A95B |
|---|---|---|
| Epoch Capabilities IndexEpoch AI, Oct 20, 2026 | #3166.32 | Not listed |
| LMArena coding chatsLMArena, Oct 15, 2026 | #10 | Not listed |
| LMArena WebDevLMArena, Oct 19, 2026 | #4 | Not listed |
| LMArena visionLMArena, Oct 19, 2026 | #9 | Not listed |
| LMArena Agent ArenaLMArena, Oct 19, 2026 | #4 | Not listed |
| SWE-Bench ProScale AI, Sep 22, 2026 | #7 | Not listed |
| MultiChallengeScale AI, Feb 3, 2026 | #1 | Not listed |
| Vals IndexVals AI, Oct 19, 2026 | #7 | Not listed |
| Terminal-Bench 4.0Vals AI, Oct 19, 2026 | #7 | Not listed |
| Vibe Code BenchVals AI, Oct 18, 2026 | #8 | Not listed |
| ProofBenchVals AI, Oct 12, 2026 | #5 | Not listed |
More comparisons