Compare / head-to-head
Qwen3 235B-A22B Instruct 2507vsQwen3 4B Thinking 2507
Qwen3 235B-A22B Instruct 2507 leads 8 of 12 shared benchmarks. Both offer a 262K-token context window.
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
8 – 4
Qwen3 235B-A22B Instruct 2507 leads
Cheaper per token
—
list price, input + output
Larger context
Tie
both 262K tokens
Providers
Qwen
same provider
Qwen3 235B-A22B Instruct 2507
Qwen · released 2025-07-21
text
Qwen3 4B Thinking 2507
Qwen · released 2025-08-06
text
- Context
- 262K
- Max out
- —
- Input /1M
- —
- Output /1M
- —
- Cached /1M
- —
- Scores
- 15 · 5 core
Quality
Benchmark matrix
12 shared · 5 only Qwen3 235B-A22B Instruct 2507 · 3 only Qwen3 4B Thinking 2507| Benchmark | Qwen3 235B-A22B Instruct 2507 | Qwen3 4B Thinking 2507 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 12 | ||||
| MMLU-Pro | 83% ↗ | 74% ↗ | +9 pt | Qwen3 235B-A22B Instruct 2507 |
| AIME25 | 70.3% ↗ | 81.3% ↗ | -11 pt | Qwen3 4B Thinking 2507 |
| Arena-Hard v2 | 79.2% ↗ | 34.9% ↗ | +44.3 pt | Qwen3 235B-A22B Instruct 2507 |
| BFCL-v3 | 70.9% ↗ | 71.2% ↗ | -0.3 pt | Qwen3 4B Thinking 2507 |
| GPQA | 77.5% ↗ | 65.8% ↗ | +11.7 pt | Qwen3 235B-A22B Instruct 2507 |
| HMMT25 | 55.4% ↗ | 55.5% ↗ | -0.1 pt | Qwen3 4B Thinking 2507 |
| IFEval | 88.7% ↗ | 87.4% ↗ | +1.3 pt | Qwen3 235B-A22B Instruct 2507 |
| LiveBench 20241125 | 75.4% ↗ | 71.8% ↗ | +3.6 pt | Qwen3 235B-A22B Instruct 2507 |
| LiveCodeBench v6 (25.02-25.05) | 51.8% ↗ | 55.2% ↗ | -3.4 pt | Qwen3 4B Thinking 2507 |
| MMLU-Redux | 93.1% ↗ | 86.1% ↗ | +7 pt | Qwen3 235B-A22B Instruct 2507 |
| SuperGPQA | 62.6% ↗ | 47.8% ↗ | +14.8 pt | Qwen3 235B-A22B Instruct 2507 |
| TAU2-Retail | 74.6% ↗ | 53.5% ↗ | +21.1 pt | Qwen3 235B-A22B Instruct 2507 |
| Only Qwen3 235B-A22B Instruct 2507 reports · 5 | ||||
| LMArena Elo | 1422.4 ↗ | not reported | — | — |
| Aider-Polyglot | 57.3% ↗ | not reported | — | — |
| ARC-AGI | 41.8% ↗ | not reported | — | — |
| SimpleQA | 54.3% ↗ | not reported | — | — |
| WeirdML (Håvard Tveit Ihle) | 38.7% ↗ | not reported | — | — |
| Only Qwen3 4B Thinking 2507 reports · 3 | ||||
| OJBench | not reported | 17.9% ↗ | — | — |
| TAU2-Airline | not reported | 58% ↗ | — | — |
| TAU2-Telecom | not reported | 27.2% ↗ | — | — |
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 Instruct 2507 minus Qwen3 4B Thinking 2507 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 Instruct 2507 | Qwen3 4B Thinking 2507 | Edge |
|---|---|---|---|
| Context window | 262K tokens | 262K tokens | Tie |
| Max output | — | — | — |
| Input price / 1M | $0.23 ↗ | — | — |
| Output price / 1M | $0.92 ↗ | — | — |
| Cached input / 1M | — | — | — |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $1.15 | — | — |
| Modalities | text | text | Tie |
| Released | 2025-07-21 | 2025-08-06 | — |
| Cited benchmark scores | 17 | 15 | — |
Reliability
Qwen status
Live from /statusMore matchups
Qwen3 235B-A22B Instruct 2507 vs …
Models sharing the most benchmarksMore matchups
Qwen3 4B Thinking 2507 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Qwen3 235B-A22B Instruct 2507 and Qwen3 4B Thinking 2507 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.