Model comparison · accepted catalogue facts 2026-11-22

Qwen2.5 72B Instruct vs Qwen3 VL 235B A22B Instruct

Qwen3 VL 235B A22B Instruct has an equal-or-higher measured score and an equal-or-lower estimated price. No known model-capability loss from Qwen2.5 72B Instruct; endpoint and task checks still apply. The reported score intervals do not overlap; this is not a task-specific success guarantee.

Both models belong to the Alibaba Qwen family. Inclusion does not recommend either model. This is a direct pair comparison; global frontier membership is not its verdict.

Qwen2.5 72B Instruct

LMArena general · Elo · higher is better

Scale starts at 1040 Elo

1269.1

Balanced effective USD / 1M tokens · lower is better

$0.370/M

Qwen3 VL 235B A22B Instruct

LMArena general · Elo · higher is better

Scale starts at 1040 Elo

1420.0

Balanced effective USD / 1M tokens · lower is better

$0.370/M

In-Depth Analysis

Qwen2.5 72B Instruct vs Qwen3 VL 235B A22B Instruct: Measured Technical Breakdown

Comprehensive comparison across measured human-preference evaluation, effective production token costs, architectural limits, and migration trade-offs.

Measured Capability (LMArena)
Qwen3 VL 235B A22B Instruct leads by 150.9 Elo

Qwen2.5 72B Instruct: 1269.1 ±4.06 · Qwen3 VL 235B A22B Instruct: 1420 ±6.51 Statistically significant preference separation

Cost Comparison Across Canonical Production Workloads

Rates represent effective cost per 1M tokens under standard token blends. Cached prompt reads are factored according to workload context reuse.

Workload & RatioQwen2.5 72B InstructQwen3 VL 235B A22B InstructCost DifferenceAdvantage
Balanced 3 tokens in per 1 out$0.370 /1M tokens$0.370 /1M tokens$0.0000 Same price
Summarise Huge input, tiny output$0.362 /1M tokens$0.208 /1M tokens$0.154 (42% diff)Qwen3 VL 235B A22B Instruct costs less
Chat Long cached system prompt$0.376 /1M tokens$0.445 /1M tokens$0.069 (16% diff)Qwen2.5 72B Instruct costs less
Code gen Output dominates$0.384 /1M tokens$0.601 /1M tokens$0.217 (36% diff)Qwen2.5 72B Instruct costs less
Agentic Context replayed each turn$0.366 /1M tokens$0.248 /1M tokens$0.118 (32% diff)Qwen3 VL 235B A22B Instruct costs less

Architectural & Operational Specifications

SpecificationQwen2.5 72B InstructQwen3 VL 235B A22B InstructOperational Note
Context window32,768 tokens262,144 tokensQwen3 VL 235B A22B Instruct has a 8.0× larger context window
Max output generation16,384 tokens32,768 tokensQwen3 VL 235B A22B Instruct allows longer single-turn completions
Base input price$0.360 / 1M tokens$0.200 / 1M tokensQwen3 VL 235B A22B Instruct input rate is cheaper
Base output price$0.400 / 1M tokens$0.880 / 1M tokensQwen2.5 72B Instruct output rate is cheaper
Cached input priceNot listed / standard rate$0.110 / 1M tokens—
Reasoning & thinking tokensStandard generationStandard generation—
Input modalitiestexttext, image—
Tool calling / function useSupportedSupported—
Open weights & licenceProprietary (API-only)Proprietary (API-only)—
Provider & reseller marketQwen · 2 provider offersQwen · 3 provider offers (1.05× spread)—

When to Choose Qwen2.5 72B Instruct

  • Direct integration with Qwen tooling and API features.
Risks when moving from Qwen3 VL 235B A22B Instruct → Qwen2.5 72B Instruct:
  • Context window: 262,144 → 32,768 tokens.
  • Output limit: 32,768 → 16,384 tokens.
  • image input is not supported by the candidate.

When to Choose Qwen3 VL 235B A22B Instruct

  • Higher measured general capability (1420 Elo on human evaluation).
  • Larger context window (262,144 tokens) for long-document or repo-level context.
  • Higher maximum completion limit (32,768 tokens).
  • Broader multimodal input support: text, image.
  • Direct integration with Qwen tooling and API features.

Compare · model decision

Choose two or three models

Inspect the differences, estimate your workload and keep the evidence. All standard catalogue models are available, including unrated ones.

Catalogue facts
2026-11-22

2 / 3 selected · 0 requirements
Open public comparison link
Public workload and requirements

These settings are included in the share link. The fixed-request scenario and private evaluation data below are excluded.

A ↔ B

Qwen3 VL 235B A22B Instruct has an equal-or-higher measured score and an equal-or-lower estimated price. No known model-capability loss from Qwen2.5 72B Instruct; endpoint and task checks still apply. The reported score intervals do not overlap; this is not a task-specific success guarantee.

Estimated blend: 75% input, 25% output; 0% cached input. Base context tier. Prompt length selects the billing tier. An unlisted cache-read rate uses that tier’s full input rate in this blend; missing required input or output rates prevent an estimate. The fixed-request estimate below validates actual context and output limits and requires a published cache-read rate when cache hits are assumed.

What changes between these models

Quality first, then estimated price. LMArena scores are human-preference measurements, not a benchmark of your application. A missing score is unrated.

  • Qwen2.5 72B Instruct · Meets the stated model requirements
  • Qwen3 VL 235B A22B Instruct · Meets the stated model requirements
Material differences and unknown evidence. Deltas compare each model with model A. Unknown facts remain visible.
Measured factA Qwen2.5 72B InstructB Qwen3 VL 235B A22B Instruct
LMArena general score (Elo)1,269.11,420Δ vs A: +150.9 Elo
Score interval half-width (± Elo)4.066.51Δ vs A: +2.45 Elo
LMArena document score (Elo)UnknownUnknown
Base input USD / 1M tokens$0.36$0.2Δ vs A: -0.16 USD/M
Base output USD / 1M tokens$0.4$0.88Δ vs A: +0.48 USD/M
Base cached-input USD / 1M tokensUnknown$0.11
Context window (tokens)32,768262,144Δ vs A: +229,376 tokens
Output limit (tokens)16,38432,768Δ vs A: +16,384 tokens
Input modalitiestexttext, image
Open weightsUnknownUnknown
Recorded licenceUnknownUnknown
Distinct sellers in the catalogue13Δ vs A: +2
Recorded retirement dateUnknownUnknown
7 shared measured facts
Shared measured facts. Deltas compare each model with model A. Unknown facts remain visible.
Measured factA Qwen2.5 72B InstructB Qwen3 VL 235B A22B Instruct
Benchmark reasoning effortdefaultdefault
Estimated effective USD / 1M tokens$0.37$0.37Δ vs A: 0 USD/M
Tool useSupportedSupportedΔ vs A: 0
ReasoningNot supportedNot supportedΔ vs A: 0
Model providerQwenQwen
Serving precision differs across offersNot supportedNot supportedΔ vs A: 0
Retirement evidenceNone announced; not a guaranteeNone announced; not a guarantee

Provider counts and precision flags describe catalogue offers. They do not establish the precision, availability, region, tool behaviour or limits of your chosen endpoint. Open each model’s provider table before switching.

Complete price ladders and billing conditions

Qwen2.5 72B Instruct

Rates below are USD per 1M tokens. A threshold selects that whole input/output rate pair.

Input-token thresholdInputOutputCached input
Base, up to and including first threshold$0.36$0.4Unknown

Qwen3 VL 235B A22B Instruct

Rates below are USD per 1M tokens. A threshold selects that whole input/output rate pair.

Input-token thresholdInputOutputCached input
Base, up to and including first threshold$0.2$0.88$0.11
Migration preflight · compare each candidate with model A

Keeping Qwen2.5 72B Instruct is a valid decision. No pairwise comparison certifies a drop-in replacement.

Qwen2.5 72B Instruct → Qwen3 VL 235B A22B Instruct

No known model-capability loss

  • Provider endpoint, serving precision, availability and task success need your own verification. Catalogue model capabilities do not certify an endpoint.
Check alternatives to Qwen2.5 72B Instruct

Estimate a fixed request shape

Private to this tab unless you save locally or explicitly export private details. These are estimates, not observed invoices. Cross-model token counts and task success are assumptions.

0.1 extra attempts means 10 extra full attempts per 100 requests. Cache writes, tool charges, taxes and unlisted charges are excluded.

  • Qwen2.5 72B InstructEstimated total: $0.92 1,000 full attempts · base rates
  • Qwen3 VL 235B A22B InstructEstimated total: $0.84 1,000 full attempts · base rates
Prompt and output sensitivity · sampled scenarios

Every point recalculates the entire request shape. Tier-boundary samples include one token below, at and above each threshold. Ordering between samples is not guaranteed. Lowest known cost excludes unknown estimates and is not a replacement recommendation.

Vary input; hold output, requests, cache and retries fixed.
Input tokensQwen2.5 72B InstructQwen3 VL 235B A22B InstructCoverage
2,000$0.92$0.84Lowest knownAll computed
4,096$1.67456$1.2592Lowest knownAll computed
32,768ineligibleInput plus output exceeds the published context window.$6.9936Lowest knownIncomplete
131,072ineligibleInput plus output exceeds the published context window.$26.6544Lowest knownIncomplete
Vary output; hold input, requests, cache and retries fixed.
Output tokensQwen2.5 72B InstructQwen3 VL 235B A22B Instruct
128$0.7712$0.51264
2,048$1.5392$2.20224
8,192$3.9968$7.60896

Bring your observed evaluation results

JSON is parsed locally. Prompts, filenames, individual responses and extra fields are not retained. Import only aggregate measurements for selected models. Task/version labels do not establish that two evaluations used the same dataset.

Expected format
[{"model":"provider/exact-model-id","task":"support-answer","version":"rubric-1","attempts":100,"successes":84,"costUsd":2.4,"latencyMs":950}]

Use the exact catalogue ID. Replace the example values with observed measurements. Omit costUsd or latencyMs when they are unknown. No public score is inferred from these observations.

Keep the decision and its evidence

A portable record retains selected facts, complete tier ladders, public requirements and the publication manifest. It also embeds the full permitted public catalogue to verify the input artifact checksum. Files are larger for this reason (up to 2 MB); local storage is limited to 12 records. Team review is local to this browser; there is no hosted collaboration.

Saving is opt-in and includes private review fields. Shared comparison links never include them. Version 1 Check references remain separate and are not overwritten or guessed into this format.

Saved local decisions · 0

No version 2 decision records are stored in this browser.

Questions this page answers

Is Qwen2.5 72B Instruct a better-and-cheaper replacement for Qwen3 VL 235B A22B Instruct?

Qwen3 VL 235B A22B Instruct has an equal-or-higher measured score and an equal-or-lower estimated price. No known model-capability loss from Qwen2.5 72B Instruct; endpoint and task checks still apply. The reported score intervals do not overlap; this is not a task-specific success guarantee.

Which model is higher rated: Qwen2.5 72B Instruct or Qwen3 VL 235B A22B Instruct?

Qwen3 VL 235B A22B Instruct scores higher on LMArena general (1420.0 ±6.5) compared to Qwen2.5 72B Instruct (1269.1 ±4.1), a margin of 150.9 Elo points. The measured confidence intervals do not overlap, demonstrating a statistically significant preference lead on human evaluations.

Which model is more cost-effective: Qwen2.5 72B Instruct or Qwen3 VL 235B A22B Instruct?

Both Qwen2.5 72B Instruct and Qwen3 VL 235B A22B Instruct share an identical effective price of $0.370 per 1M tokens on a balanced workload.

How do their context windows and generation limits compare?

Qwen2.5 72B Instruct provides a 32,768-token context window with a maximum output limit of 16,384 tokens. Qwen3 VL 235B A22B Instruct provides a 262,144-token context window with a maximum output of 32,768 tokens. Qwen3 VL 235B A22B Instruct offers a 8.0× larger context capacity, making it suitable for larger repositories and multi-document workflows.

What capabilities are lost when switching between Qwen2.5 72B Instruct and Qwen3 VL 235B A22B Instruct?

Switching from Qwen2.5 72B Instruct to Qwen3 VL 235B A22B Instruct causes no detected architectural capability loss. Conversely, moving from Qwen3 VL 235B A22B Instruct to Qwen2.5 72B Instruct involves: Context window: 262,144 → 32,768 tokens; Output limit: 32,768 → 16,384 tokens; image input is not supported by the candidate. Run test suites on your specific prompt templates and tool declarations before cutting over production traffic.

Can Qwen2.5 72B Instruct or Qwen3 VL 235B A22B Instruct be self-hosted or deployed privately?

Both Qwen2.5 72B Instruct and Qwen3 VL 235B A22B Instruct are proprietary cloud models offered via managed API endpoints. Qwen2.5 72B Instruct is served by Qwen and 2 catalogue offers, while Qwen3 VL 235B A22B Instruct is served by Qwen and 3 catalogue offers.

What is included in this comparison?

Accepted model capabilities, context and output limits, complete price-tier ladders, independent LMArena measurements and missing evidence. Provider endpoint precision and application success still need verification.

Can I compare a different workload or keep a decision record?

Yes. Change the public workload, prompt length and requirements in the comparison workspace. Fixed-request scenarios, aggregate evaluations and review notes stay local unless explicitly exported.

Evidence & Ask