Is Qwen3 235B A22B Thinking 2507 a good deal?
Whether anything in this catalogue beats Qwen3 235B A22B Thinking 2507 on both quality and price, and what you give up if it does. A computation on the current catalogue, not an opinion.
12 undominated of 144 · Nov 22, 2026
As of Nov 22, 2026, Qwen3 235B A22B Thinking 2507 is dominated for Balanced on LMArena. MiMo-V2.6-Pro scores 77.3 higher and costs 55% less, with a covering envelope. 12 of 144 rated, priced standard models are undominated.
Inspect model evidence Compare differences & requirements
MiMo-V2.6-Pro is both better and cheaper than Qwen3 235B A22B Thinking 2507: 77.3 points higher on LMArena and 55% less per million tokens, $0.673 cheaper at this mix.
LMArena Elo · higher is better
Scale starts at 1040 Elo
1415.4
Effective $/M · Balanced · lower is better
$1.21/M
LMArena Elo · higher is better
Scale starts at 1040 Elo
1492.7
Effective $/M · Balanced · lower is better
$0.540/M
Qwen3 235B A22B Thinking 2507 takes text, returns up to 16,384 tokens from a 128,000-token context, and is listed by 1 seller.
Compared against 144 rated, priced models on this lens: 34 models dominate it and give up nothing, 3 more dominate it but give something up. Qwen3 235B A22B Thinking 2507 scores 1415.4 at $1.21 per million tokens for this mix.
Envelope-safe replacements
Each row scores at least as high, costs no more, and covers this model’s context, output, modalities, tools, and reasoning. A cheaper narrower model is not listed here.
| Model | LMArena | Effective $/M | You save |
|---|---|---|---|
| MiMo-V2.6-Pro | 1492.7 +77.3 | $0.540/M | 55% |
| GLM 5.3 | 1475.4 +60.0 | $1.000/M | 18% |
| GLM 5.3 Flash | 1471.4 +56.0 | $0.148/M | 88% |
| GLM 5.2 | 1469.5 +54.1 | $0.977/M | 19% |
| MiMo-V2.5-Pro | 1463.9 +48.5 | $0.810/M | 33% |
| DeepSeek V4.1 Flash | 1463.0 +47.6 | $0.227/M | 81% |
| Qwen3.7 Plus | 1454.1 +38.7 | $0.560/M | 54% |
| MiMo-V2.6-Flash | 1449.8 +34.4 | $0.175/M | 86% |
| GLM 5 | 1446.1 +30.7 | $1.09/M | 11% |
| DeepSeek V4 Pro 0423 | 1444.7 +29.3 | $1.10/M | 9% |
| Kimi K2.5 | 1444.7 +29.3 | $1.20/M | 1% |
| Gemma 4 31B | 1442.5 +27.1 | $0.205/M | 83% |
| GLM 4.6 | 1440.3 +24.9 | $0.760/M | 37% |
| Hy3 | 1439.0 +23.6 | $0.231/M | 81% |
| Qwen3.5 397B A17B | 1437.6 +22.2 | $0.878/M | 28% |
| Qwen3.8 27B | 1437.4 +22.0 | $0.956/M | 21% |
| Qwen3.6 Plus | 1436.2 +20.8 | $0.731/M | 40% |
| GLM 4.7 | 1435.4 +20.0 | $0.738/M | 39% |
| Gemini 3.5 Flash Lite | 1435.0 +19.6 | $0.850/M | 30% |
| Gemma 4 26B A4B | 1434.2 +18.8 | $0.123/M | 90% |
| DeepSeek V4 Flash 0423 | 1431.9 +16.5 | $0.113/M | 91% |
| GPT-5.6 Luna | 1430.9 +15.5 | $0.450/M | 63% |
| GLM 4.5 | 1430.2 +14.8 | $1.00/M | 18% |
| MiMo-V2.5 | 1427.7 +12.3 | $0.800/M | 34% |
| R1 0528 | 1427.7 +12.3 | $0.920/M | 24% |
| MiniMax M3 | 1426.5 +11.1 | $0.525/M | 57% |
| DeepSeek V3.2 | 1424.6 +9.2 | $0.394/M | 67% |
| Beam 1.0 501B A23B | 1424.3 +8.9 | $0.600/M | 51% |
| DeepSeek V3.2 Exp | 1422.3 +6.9 | $0.305/M | 75% |
| DeepSeek V3.1 | 1419.8 +4.4 | $0.453/M | 63% |
| Qwen3.5-122B-A10B | 1417.5 +2.1 | $0.715/M | 41% |
| DeepSeek V3.1 Terminus | 1417.3 +1.9 | $0.453/M | 63% |
| Gemini 2.5 Flash | 1416.9 +1.5 | $0.850/M | 30% |
| Gemini 3.1 Flash Lite Preview | 1416.2 +0.8 | $0.563/M | 54% |
Higher score, lower price, named losses
Not a drop-in. The loss is why these are not a recommendation.
| Model | LMArena | Effective $/M | You give up |
|---|---|---|---|
| Qwen3 VL 235B A22B Instruct | 1420.0 | $0.370/M | reasoning |
| Qwen3 235B A22B Instruct 2507 | 1419.2 | $0.205/M | reasoning |
| Qwen3 Next 80B A3B Instruct | 1417.2 | $0.343/M | reasoning |
The link keeps the model and mix, using the current catalogue. Monthly spend and switching cost stay in this browser tab and are left out of the link.
Saved decision references
Save the model, workload, catalogue date and capability-preservation rule in this browser. Spend, switching cost and bill contents are not saved. No account or notifications.
Constraint: replacements must preserve the model’s capabilities; any losses remain named trade-offs.
Historical decisions cannot be fully replayed from saved references: past prices, scores and capability evidence are not stored.