Across 7 comparable benchmarks, Qwen3-235B-A22B has better reported results on 4 and worse on 3.
Moderate evidence7 of 7 shared results countedHow we compare
Price basis: Listed rates vs listed rates. The price section below shows the applicable periods.
| Fact | Qwen3-235B-A22B | MiniCPM5-2B |
|---|---|---|
| Context window | 128K tokens | 128K tokens |
| Input / 1M | — | — |
| Output / 1M | — | — |
| Blended (4:1) | — | — |
| Status | Deprecated | Available |
| Released | April 28, 2025 | September 7, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 4 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
Split 1–1 of 2 counted
Split 1–1 of 2 counted
Qwen3-235B-A22B ahead on 1 of 1 counted
No shared comparable results for Abstract reasoning, Computer & tool use, Research & search, Long context, Multimodal, Professional work & productivity, Cybersecurity.
Published scores, with matching modes where available. Gaps show Qwen3-235B-A22B relative to MiniCPM5-2B. Reading the results
| Benchmark | Qwen3-235B-A22B | MiniCPM5-2B | Gap |
|---|---|---|---|
| LiveCodeBenchAlgorithmic Coding · Pass @K | 70.7Thinking · no tools | 69.1≠Reported best | 1.02×#52 vs #57 of 113 |
| SWE-bench VerifiedRepository Engineering · Accuracy | 34.4Normal · no tools | 46.4≠Reported best | −12.0 pts#102 vs #97 of 108 |
| AIME2025Mathematics · Accuracy | 81.5Thinking · no tools | 86.5≠Reported best | −5.0 pts#53 vs #44 of 86 |
| MATH-500Mathematics · Accuracy | 98.0Thinking · no tools | 94.6≠Reported best | +3.4 pts#16 vs #33 of 49 |
| GPQA DiamondScientific Reasoning · Accuracy | 71.1Normal · no tools | 70.2≠Reported best | +0.9 pts#135 vs #139 of 200 |
| HLEKnowledge Exams · Accuracy | 7.6Normal · no tools | 8.9≠Reported best | −1.3 pts#135 vs #132 of 151 |
| MMLU-ProKnowledge Exams · Accuracy | 72.9Normal · no tools | 70.8≠Reported best | +2.1 pts#86 vs #91 of 153 |
Algorithmic Coding · Pass @K
Repository Engineering · Accuracy
Scientific Reasoning · Accuracy
A dash means no result is available.
| Benchmark | Qwen3-235B-A22B | MiniCPM5-2B |
|---|---|---|
| AIME 2024 | 85.7 | — |
| AIME 2026 | — | 86.5 |
| CritPt | — | 0.3 |
| GSM8K | 96.4 | — |
| HMMT Feb 2026 | — | 63.8 |
| MMLU | 85.8 | — |
| SciCode | — | 26.3 |
| Aider-Polyglot | 59.6 | — |
| SWE-Bench Pro - Public | — | 14.4 |
| Terminal Bench Hard | 6.1 | — |
| Terminal-Bench 2.1 | — | 8.6 |
| BFCL-V4 | — | 66.6 |
| ECI | 139 | — |
| DROP | 88.7 | — |
| IF Bench | — | 66.3 |
| IFEval | — | 86.7 |
| LongBench v2 | — | 43.7 |
| GDP.pdf | — | 1.4 |
| τ²-Bench | 34.4 | — |
| τ²-Bench - Telecom | — | 97.1 |
| τ³-Banking | — | 20.8 |
| ARC-AGI-1 | 4.3 | — |
| BBH | 88.9 | — |
| SimpleBench | 31.0 | — |
| SimpleQA | 11.0 | — |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
Qwen3-235B-A22B: Listed text rates; pricing tier not confirmed.
MiniCPM5-2B: Listed text rates; pricing tier not confirmed.
| Price / 1M tokens | Qwen3-235B-A22B | MiniCPM5-2B | Ratio |
|---|---|---|---|
| Input | — | — | — |
| Output | — | — | — |
| Cached input | — | — | — |
| Cache write | — | — | — |
| Blended | — | — | — |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Lower prices are highlighted. The cache hit rate applies to input tokens only; models without a published cached-input rate use their normal input price for those tokens.| Attribute | Qwen3-235B-A22B | MiniCPM5-2B |
|---|---|---|
| Availability | Deprecated | Available |
| Provider | Alibaba | OpenBMB |
| Model type | Reasoning model | Reasoning model |
| First release | April 28, 2025 | September 7, 2026 |
| Context window | 128K tokens | 128K tokens |
| Max output | 16K tokens | — |
| Total parameters | 235B | 2.5B |
| Active parameters | 22B | 2.5B |
| Architecture | Mixture of Experts (MoE) | Dense |
| Documented capabilities | Reasoning, Multilingual | Reasoning, Multilingual, Coding, Function calling, Tool use, Web search, Long context |
| Input modalities | Text | Text |
| Output modalities | Text | Text |
| Weights | Open weights | Open weights |
| Licence | Apache 2.0 | Apache 2.0 |
| Commercial use | Commercial use allowed | Commercial use allowed |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used Qwen3-235B-A22B or MiniCPM5-2B? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.