Compare / head-to-head
Qwen3 235B-A22B Thinking 2507vsQwen3-VL 235B-A22B Instruct
Qwen3 235B-A22B Thinking 2507 leads 7 of 9 shared benchmarks. Qwen3-VL 235B-A22B Instruct is 1.3x cheaper per token. Qwen3 235B-A22B Thinking 2507 has the larger context window (262K tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
7 – 1
Qwen3 235B-A22B Thinking 2507 leads · 1 tied
Cheaper per token
Qwen3-VL 235B-A22B Instruct
1.3x cheaper, input + output
Larger context
Qwen3 235B-A22B Thinking 2507
262K tokens
Providers
Qwen
same provider
Qwen3 235B-A22B Thinking 2507
Qwen · released 2025-07-25
text
Qwen3-VL 235B-A22B Instruct
Qwen · released 2025-09-23
textvisionvideo
Quality
Benchmark matrix
9 shared · 14 only Qwen3 235B-A22B Thinking 2507 · 8 only Qwen3-VL 235B-A22B Instruct| Benchmark | Qwen3 235B-A22B Thinking 2507 | Qwen3-VL 235B-A22B Instruct | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 9 | ||||
| MMLU-Pro | 84.4% ↗ | 81.8% ↗ | +2.6 pt | Qwen3 235B-A22B Thinking 2507 |
| LMArena Elo | 1400 ↗ | 1413.4 ↗ | -13.4 | Qwen3-VL 235B-A22B Instruct |
| AIME25 | 92.3% ↗ | 74.7% ↗ | +17.6 pt | Qwen3 235B-A22B Thinking 2507 |
| Arena-Hard v2 | 79.7% ↗ | 77.4% ↗ | +2.3 pt | Qwen3 235B-A22B Thinking 2507 |
| BFCL-v3 | 71.9% ↗ | 67.7% ↗ | +4.2 pt | Qwen3 235B-A22B Thinking 2507 |
| HMMT25 | 83.9% ↗ | 57.4% ↗ | +26.5 pt | Qwen3 235B-A22B Thinking 2507 |
| IFEval | 87.8% ↗ | 87.8% ↗ | 0 | Tie |
| MMLU-Redux | 93.8% ↗ | 92.2% ↗ | +1.6 pt | Qwen3 235B-A22B Thinking 2507 |
| SuperGPQA | 64.9% ↗ | 60.4% ↗ | +4.5 pt | Qwen3 235B-A22B Thinking 2507 |
| Only Qwen3 235B-A22B Thinking 2507 reports · 14 | ||||
| GPQA Diamond | 80.1% ↗epoch run | not reported | — | — |
| Chess Puzzles (Epoch AI run) | 12% ↗epoch run | not reported | — | — |
| GPQA | 81.1% ↗ | not reported | — | — |
| HLE (text-only) | 18.2% ↗ | not reported | — | — |
| LiveBench 20241125 | 78.4% ↗ | not reported | — | — |
| LiveCodeBench v6 (25.02-25.05) | 74.1% ↗ | not reported | — | — |
| Mystery Game Puzzles (Epoch AI run) | 9% ↗epoch run | not reported | — | — |
| OJBench | 32.5% ↗ | not reported | — | — |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 86.7% ↗epoch run | not reported | — | — |
| SimpleQA Verified | 40.4% ↗epoch run | not reported | — | — |
| TAU2-Airline | 58% ↗ | not reported | — | — |
| TAU2-Retail | 71.9% ↗ | not reported | — | — |
| Vending-Bench 2 (Andon Labs) | -11.34 ↗ | not reported | — | — |
| WeirdML (Håvard Tveit Ihle) | 41% ↗ | not reported | — | — |
| Only Qwen3-VL 235B-A22B Instruct reports · 8 | ||||
| LCBV6 (25.02-25.05) | not reported | 54.3% ↗ | — | — |
| LMArena Vision (LMArena) | not reported | 1215.1 ↗ | — | — |
| MathVista (mini) | not reported | 84.9% ↗ | — | — |
| MMMU (val) | not reported | 78.7% ↗ | — | — |
| MMMU_Pro | not reported | 68.1% ↗ | — | — |
| ScreenSpot Pro | not reported | 62% ↗ | — | — |
| SimpleQA | not reported | 51.9% ↗ | — | — |
| VideoMME (w/o sub) | not reported | 79.2% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is Qwen3 235B-A22B Thinking 2507 minus Qwen3-VL 235B-A22B Instruct in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | Qwen3 235B-A22B Thinking 2507 | Qwen3-VL 235B-A22B Instruct | Edge |
|---|---|---|---|
| Context window | 262K tokens | 256K tokens | Qwen3 235B-A22B Thinking 2507 |
| Max output | — | — | — |
| Input price / 1M | $0.23 ↗ | $0.4 ↗ | Qwen3 235B-A22B Thinking 2507 |
| Output price / 1M | $2.3 ↗ | $1.6 ↗ | Qwen3-VL 235B-A22B Instruct |
| Cached input / 1M | — | — | — |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $2.53 | $2 | Qwen3-VL 235B-A22B Instruct |
| Modalities | text | text · vision · video | Qwen3-VL 235B-A22B Instruct |
| Released | 2025-07-25 | 2025-09-23 | — |
| Cited benchmark scores | 23 | 17 | — |
Reliability
Qwen status
Live from /statusMore matchups
Qwen3 235B-A22B Thinking 2507 vs …
Models sharing the most benchmarksMore matchups
Qwen3-VL 235B-A22B Instruct vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Qwen3 235B-A22B Thinking 2507 and Qwen3-VL 235B-A22B Instruct on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.