Qwen2-72B vs Qwen3-235B-A22B
Qwen2-72B (2024) and Qwen3-235B-A22B (2025) are compact production models from Alibaba. Qwen2-72B ships a 128k-token context window, while Qwen3-235B-A22B ships a 128k-token context window. On MMLU PRO, Qwen3-235B-A22B leads by 18.4 pts. On pricing, Qwen3-235B-A22B costs $0.09/1M input tokens versus $0.45/1M for the alternative. This comparison covers specs, pricing, capabilities, benchmarks, provider availability, and production fit. It focuses on practical selection signals rather than broad model-family marketing.
Qwen3-235B-A22B is ~400% cheaper at $0.09/1M; pay for Qwen2-72B only for provider fit.
Decision scorecard
Local evidence first| Signal | Qwen2-72B | Qwen3-235B-A22B |
|---|---|---|
| Best for | provider-routed production | provider-routed production |
| Decision fit | Coding, RAG, and Long context | Coding, RAG, and Long context |
| Context window | 128k | 128k |
| Cheapest output | $0.65/1M tokens | $0.58/1M tokens |
| Provider routes | 4 tracked | 7 tracked |
| Shared benchmarks | 2 rows | MMLU PRO leader |
Decision tradeoffs
- Local decision data tags Qwen2-72B for Coding, RAG, and Long context.
- Qwen3-235B-A22B leads the largest shared benchmark signal on MMLU PRO by 18.4 points.
- Qwen3-235B-A22B has the lower cheapest tracked output price at $0.58/1M tokens.
- Qwen3-235B-A22B has broader tracked provider coverage for fallback and procurement flexibility.
- Local decision data tags Qwen3-235B-A22B for Coding, RAG, and Long context.
Monthly cost at traffic
Estimate token spend from the cheapest tracked input and output route or tier on this page.
Qwen2-72B
$523
Cheapest tracked route/tier: DeepInfra
Qwen3-235B-A22B
$217
Cheapest tracked route/tier: Novita AI
Estimated monthly gap: $306. Batch, cache, alternate speed tiers, and negotiated pricing are excluded from this local estimate.
Switch friction
- Provider overlap exists on Fireworks AI; start route-level A/B tests there.
- Qwen3-235B-A22B is $0.07/1M tokens lower on cheapest tracked output pricing before cache, batch, or negotiated discounts.
- Provider overlap exists on Fireworks AI; start route-level A/B tests there.
- Qwen2-72B is $0.07/1M tokens higher on cheapest tracked output pricing, so quality gains need to justify the spend.
Specs
| Specification | ||
|---|---|---|
| Released | 2024-06-05 | 2025-04-29 |
| Context window | 128k | 128k |
| Parameters | 72.71B | 235B |
| Architecture | decoder only | decoder only |
| License | Apache 2.0 | Apache 2.0 |
| Knowledge cutoff | - | - |
Pricing and availability
| Pricing attribute | Qwen2-72B | Qwen3-235B-A22B |
|---|---|---|
| Input price | $0.45/1M tokens | $0.09/1M tokens |
| Output price | $0.65/1M tokens | $0.58/1M tokens |
| Providers |
Capabilities
| Capability | Qwen2-72B | Qwen3-235B-A22B |
|---|---|---|
| Vision | No | No |
| Multimodal | No | No |
| Reasoning | No | No |
| Function calling | No | No |
| Tool use | No | No |
| Structured outputs | Yes | Yes |
| Code execution | No | No |
| IDE integration | No | No |
| Computer use | No | No |
| Parallel agents | No | No |
Benchmarks
| Benchmark | Qwen2-72B | Qwen3-235B-A22B |
|---|---|---|
| MMLU PRO | 64.4 | 82.8 |
| HumanEval | 67.1 | 92.7 |
Deep dive
On shared benchmark coverage, MMLU PRO has Qwen2-72B at 64.4 and Qwen3-235B-A22B at 82.8, with Qwen3-235B-A22B ahead by 18.4 points; HumanEval has Qwen2-72B at 67.1 and Qwen3-235B-A22B at 92.7, with Qwen3-235B-A22B ahead by 25.6 points. The largest visible gap is 25.6 points on HumanEval, which matters most when that benchmark mirrors your workload. Treat isolated benchmark wins as directional, because provider routing, prompt style, and tool access can move real application results.
The capability footprint is close: both models cover structured outputs. That makes context budget, benchmark fit, and provider maturity more important than a simple checklist. If your application depends on one integration detail, verify it against the provider route you plan to use, not just the base model listing.
For cost, Qwen2-72B lists $0.45/1M input and $0.65/1M output tokens on the cheapest tracked provider, while Qwen3-235B-A22B lists $0.09/1M input and $0.58/1M output tokens on the cheapest tracked provider. A 70/30 input-output blend puts Qwen3-235B-A22B lower by about $0.27 per million blended tokens. Availability is 4 providers versus 7, so concentration risk also matters.
Choose Qwen2-72B when provider fit are central to the workload. Choose Qwen3-235B-A22B when provider fit, lower input-token cost, and broader provider choice are more important. For production, rerun your own prompts through the exact provider, region, and tool stack you plan to ship.
FAQ
Which has a larger context window, Qwen2-72B or Qwen3-235B-A22B?
Qwen2-72B supports 128k tokens, while Qwen3-235B-A22B supports 128k tokens. That gap matters most for long documents, large codebases, retrieval-heavy agents, and conversations where earlier context must remain visible. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Which is cheaper, Qwen2-72B or Qwen3-235B-A22B?
Qwen3-235B-A22B is cheaper on tracked token pricing. Qwen2-72B costs $0.45/1M input and $0.65/1M output tokens. Qwen3-235B-A22B costs $0.09/1M input and $0.58/1M output tokens. Provider discounts or batch pricing can still change the final bill.
Is Qwen2-72B or Qwen3-235B-A22B open source?
Qwen2-72B is listed under Apache 2.0. Qwen3-235B-A22B is listed under Apache 2.0. License labels affect whether you can self-host, redistribute weights, or rely only on hosted APIs, so confirm the upstream license before deployment.
Which is better for structured outputs, Qwen2-72B or Qwen3-235B-A22B?
Both Qwen2-72B and Qwen3-235B-A22B expose structured outputs. The better choice depends on benchmark fit, context budget, pricing, and whether your provider route exposes the same capability surface. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Where can I run Qwen2-72B and Qwen3-235B-A22B?
Qwen2-72B is available on Fireworks AI, DeepInfra, Together AI, and Microsoft Foundry. Qwen3-235B-A22B is available on Fireworks AI, AWS Bedrock, OpenRouter, Venice AI, and Novita AI. Provider coverage can affect latency, region availability, compliance posture, and fallback options.
When should I pick Qwen2-72B over Qwen3-235B-A22B?
Qwen3-235B-A22B is ~400% cheaper at $0.09/1M; pay for Qwen2-72B only for provider fit. If your workload also depends on provider fit, start with Qwen2-72B; if it depends on provider fit, run the same evaluation with Qwen3-235B-A22B.
Continue comparing
Last reviewed: 2026-05-22. Data sourced from public model cards and provider documentation.