Qwen2.5-72B-Instruct vs Qwen3.5-397B-A17B
Qwen2.5-72B-Instruct (2024) and Qwen3.5-397B-A17B (2026) are frontier reasoning models from Alibaba. Qwen2.5-72B-Instruct ships a 128k-token context window, while Qwen3.5-397B-A17B ships a 262k-token context window. On Google-Proof Q&A, Qwen3.5-397B-A17B leads by 50.9 pts. On pricing, Qwen2.5-72B-Instruct costs $0.18/1M input tokens versus $0.39/1M for the alternative. This comparison covers specs, pricing, API access, capabilities, benchmarks, input and output token costs, and production fit for coding and agent workloads.
Qwen2.5-72B-Instruct is ~117% cheaper at $0.18/1M; pay for Qwen3.5-397B-A17B only for reasoning depth.
Decision scorecard
Local evidence first| Signal | Qwen2.5-72B-Instruct | Qwen3.5-397B-A17B |
|---|---|---|
| Best for | provider-routed production | reasoning-heavy apps, multimodal apps, and tool-calling agents |
| Decision fit | Coding, RAG, and Long context | Coding, RAG, and Agents |
| Context window | 128k | 262k |
| Cheapest output | $0.54/1M tokens | $2.34/1M tokens |
| Provider routes | 7 tracked | 4 tracked |
| Shared benchmarks | 1 rows | Google-Proof Q&A leader |
Decision tradeoffs
- Qwen2.5-72B-Instruct has the lower cheapest tracked output price at $0.54/1M tokens.
- Qwen2.5-72B-Instruct has broader tracked provider coverage for fallback and procurement flexibility.
- Local decision data tags Qwen2.5-72B-Instruct for Coding, RAG, and Long context.
- Qwen3.5-397B-A17B holds a shared-benchmark lead on Google-Proof Q&A, ahead by 50.9 points.
- Qwen3.5-397B-A17B has the larger context window for long prompts, retrieval packs, or transcript analysis.
- Qwen3.5-397B-A17B uniquely exposes Vision, Multimodal, and Reasoning in local model data.
- Local decision data tags Qwen3.5-397B-A17B for Coding, RAG, and Agents.
Monthly cost at traffic
Estimate token spend from the cheapest tracked input and output route or tier on this page.
Qwen2.5-72B-Instruct
$279
Cheapest tracked route/tier: Chutes AI
Qwen3.5-397B-A17B
$897
Cheapest tracked route/tier: OpenRouter
Estimated monthly gap: $618. Batch, cache, alternate speed tiers, and negotiated pricing are excluded from this local estimate.
Switch friction
- Provider overlap exists on OpenRouter and Novita AI; start route-level A/B tests there.
- Qwen3.5-397B-A17B is $1.80/1M tokens higher on cheapest tracked output pricing, so quality gains need to justify the spend.
- Qwen3.5-397B-A17B adds Vision, Multimodal, and Reasoning in local capability data.
- Provider overlap exists on OpenRouter and Novita AI; start route-level A/B tests there.
- Qwen2.5-72B-Instruct is $1.80/1M tokens lower on cheapest tracked output pricing before cache, batch, or negotiated discounts.
- Check replacement coverage for Vision, Multimodal, and Reasoning before moving production traffic.
Specs
| Specification | ||
|---|---|---|
| Released | 2024-06-07 | 2026-02-16 |
| Context window | 128k | 262k |
| Parameters | 72.7B | 397B |
| Architecture | decoder only | MoE |
| License | Apache 2.0(OSI) | Apache 2.0(OSI) |
| Openness | Open source | Open source |
| Commercial use | Commercial use allowed | Commercial use allowed |
| Knowledge cutoff | - | - |
Pricing and availability
| Pricing attribute | Qwen2.5-72B-Instruct | Qwen3.5-397B-A17B |
|---|---|---|
| Input price | $0.18/1M tokens | $0.39/1M tokens |
| Output price | $0.54/1M tokens | $2.34/1M tokens |
| Providers |
Capabilities
| Capability | Qwen2.5-72B-Instruct | Qwen3.5-397B-A17B |
|---|---|---|
| Vision | No | Yes |
| Multimodal | No | Yes |
| Reasoning | No | Yes |
| Function calling | No | Yes |
| Tool use | No | Yes |
| Structured outputs | Yes | Yes |
| Code execution | No | No |
| IDE integration | No | No |
| Computer use | No | No |
| Parallel agents | No | No |
Benchmarks
| Benchmark | Qwen2.5-72B-Instruct | Qwen3.5-397B-A17B |
|---|---|---|
| Google-Proof Q&A | 38.4 | 89.3 |
Deep dive
On shared benchmark coverage, Google-Proof Q&A has Qwen2.5-72B-Instruct at 38.4 and Qwen3.5-397B-A17B at 89.3, with Qwen3.5-397B-A17B ahead by 50.9 points. The largest visible gap is 50.9 points on Google-Proof Q&A, 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 differs most on vision: Qwen3.5-397B-A17B, multimodal input: Qwen3.5-397B-A17B, reasoning mode: Qwen3.5-397B-A17B, function calling: Qwen3.5-397B-A17B, and tool use: Qwen3.5-397B-A17B. Both models share structured outputs, so the practical split is not just feature count. Use those differences to decide whether the page is about raw model quality, agentic coding support, multimodal ingestion, or predictable structured API behavior.
For cost, Qwen2.5-72B-Instruct lists $0.18/1M input and $0.54/1M output tokens on the cheapest tracked provider, while Qwen3.5-397B-A17B lists $0.39/1M input and $2.34/1M output tokens on the cheapest tracked provider. A 70/30 input-output blend puts Qwen2.5-72B-Instruct lower by about $0.69 per million blended tokens. Availability is 7 providers versus 4, so concentration risk also matters.
Choose Qwen2.5-72B-Instruct when provider fit, lower input-token cost, and broader provider choice are central to the workload. Choose Qwen3.5-397B-A17B when reasoning depth and larger context windows 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, Qwen2.5-72B-Instruct or Qwen3.5-397B-A17B?
Qwen3.5-397B-A17B supports 262k tokens, while Qwen2.5-72B-Instruct 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, Qwen2.5-72B-Instruct or Qwen3.5-397B-A17B?
Qwen2.5-72B-Instruct is cheaper on tracked token pricing. Qwen2.5-72B-Instruct costs $0.18/1M input and $0.54/1M output tokens. Qwen3.5-397B-A17B costs $0.39/1M input and $2.34/1M output tokens. Provider discounts or batch pricing can still change the final bill.
Is Qwen2.5-72B-Instruct or Qwen3.5-397B-A17B open source?
Qwen2.5-72B-Instruct is listed under Apache 2.0. Qwen3.5-397B-A17B 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 vision, Qwen2.5-72B-Instruct or Qwen3.5-397B-A17B?
Qwen3.5-397B-A17B has the clearer documented vision signal in this comparison. If vision is mission-critical, validate it against the provider endpoint because model-level support and API-level exposure can differ. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Which is better for multimodal input, Qwen2.5-72B-Instruct or Qwen3.5-397B-A17B?
Qwen3.5-397B-A17B has the clearer documented multimodal input signal in this comparison. If multimodal input is mission-critical, validate it against the provider endpoint because model-level support and API-level exposure can differ.
Where can I run Qwen2.5-72B-Instruct and Qwen3.5-397B-A17B?
Qwen2.5-72B-Instruct is available on DeepInfra, OpenRouter, Fireworks AI, Novita AI, and Chutes AI. Qwen3.5-397B-A17B is available on OpenRouter, Together AI, Alibaba Cloud PAI-EAS, and Novita AI. Provider coverage can affect latency, region availability, compliance posture, and fallback options.
Continue comparing
Last reviewed: 2026-06-04. Data sourced from public model cards and provider documentation.