Model comparison · accepted catalogue facts 2026-11-22

Qwen3.5 397B A17B vs Qwen3.8 27B

Qwen3.5 397B A17B has an equal-or-higher measured score and an equal-or-lower estimated price. Switching from Qwen3.8 27B loses capability or fails a requirement. Review the migration checks. The reported score intervals overlap; this is not an established quality improvement.

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.

Qwen3.5 397B A17B

LMArena general · Elo · higher is better

Scale starts at 1040 Elo

1437.6

Balanced effective USD / 1M tokens · lower is better

$0.878/M

Qwen3.8 27B

LMArena general · Elo · higher is better

Scale starts at 1040 Elo

1437.4

Balanced effective USD / 1M tokens · lower is better

$0.956/M

In-Depth Analysis

Qwen3.5 397B A17B vs Qwen3.8 27B: Measured Technical Breakdown

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

Measured Capability (LMArena)
Qwen3.5 397B A17B leads by 0.2 Elo

Qwen3.5 397B A17B: 1437.6 ±3.03 · Qwen3.8 27B: 1437.4 ±4.44 Intervals overlap — not a guaranteed quality advantage

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 & RatioQwen3.5 397B A17BQwen3.8 27BCost DifferenceAdvantage
Balanced 3 tokens in per 1 out$0.878 /1M tokens$0.956 /1M tokens$0.079 (8% diff)Qwen3.5 397B A17B costs less
Summarise Huge input, tiny output$0.488 /1M tokens$0.434 /1M tokens$0.053 (11% diff)Qwen3.8 27B costs less
Chat Long cached system prompt$1.17 /1M tokens$1.17 /1M tokens$0.0030 Qwen3.5 397B A17B costs less
Code gen Output dominates$1.56 /1M tokens$1.67 /1M tokens$0.113 (7% diff)Qwen3.5 397B A17B costs less
Agentic Context replayed each turn$0.683 /1M tokens$0.541 /1M tokens$0.141 (21% diff)Qwen3.8 27B costs less

Architectural & Operational Specifications

SpecificationQwen3.5 397B A17BQwen3.8 27BOperational Note
Context window262,144 tokens1,024,000 tokensQwen3.8 27B has a 3.9× larger context window
Max output generation81,920 tokens131,072 tokensQwen3.8 27B allows longer single-turn completions
Base input price$0.390 / 1M tokens$0.425 / 1M tokensQwen3.5 397B A17B input rate is cheaper
Base output price$2.34 / 1M tokens$2.55 / 1M tokensQwen3.5 397B A17B output rate is cheaper
Cached input priceNot listed / standard rate$0.085 / 1M tokens—
Reasoning & thinking tokensSupported; standard ratesSupported; standard rates—
Input modalitiestext, image, videotext, image, video—
Tool calling / function useSupportedSupported—
Open weights & licenceProprietary (API-only)Open weights (apache-2.0)Qwen3.8 27B allows independent hosting or local deployment
Provider & reseller marketQwen · 8 provider offers (2.08× spread)Qwen · 23 provider offers (28.29× spread)—

When to Choose Qwen3.5 397B A17B

  • Higher measured general capability (1437.6 Elo on human evaluation).
  • Direct integration with Qwen tooling and API features.
Risks when moving from Qwen3.8 27B → Qwen3.5 397B A17B:
  • Context window: 1,024,000 → 262,144 tokens.
  • Output limit: 131,072 → 81,920 tokens.

When to Choose Qwen3.8 27B

  • Larger context window (1,024,000 tokens) for long-document or repo-level context.
  • Higher maximum completion limit (131,072 tokens).
  • Open weights availability allows private self-hosting and on-premise execution.
  • 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

Qwen3.5 397B A17B has an equal-or-higher measured score and an equal-or-lower estimated price. Switching from Qwen3.8 27B loses capability or fails a requirement. Review the migration checks. The reported score intervals overlap; this is not an established quality improvement.

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.

  • Qwen3.5 397B A17B · Meets the stated model requirements
  • Qwen3.8 27B · Meets the stated model requirements
Material differences and unknown evidence. Deltas compare each model with model A. Unknown facts remain visible.
Measured factA Qwen3.5 397B A17BB Qwen3.8 27B
LMArena general score (Elo)1,437.61,437.4Δ vs A: -0.2 Elo
Score interval half-width (± Elo)3.034.44Δ vs A: +1.41 Elo
LMArena document score (Elo)UnknownUnknown
Estimated effective USD / 1M tokens$0.8775$0.95625Δ vs A: +0.07875 USD/M
Base input USD / 1M tokens$0.39$0.425Δ vs A: +0.035 USD/M
Base output USD / 1M tokens$2.34$2.55Δ vs A: +0.21 USD/M
Base cached-input USD / 1M tokensUnknown$0.085
Context window (tokens)262,1441,024,000Δ vs A: +761,856 tokens
Output limit (tokens)81,920131,072Δ vs A: +49,152 tokens
Open weightsUnknownSupported
Recorded licenceUnknownapache-2.0
Distinct sellers in the catalogue623Δ vs A: +17
Recorded retirement dateUnknownUnknown
7 shared measured facts
Shared measured facts. Deltas compare each model with model A. Unknown facts remain visible.
Measured factA Qwen3.5 397B A17BB Qwen3.8 27B
Benchmark reasoning effortdefaultdefault
Input modalitiestext, image, videotext, image, video
Tool useSupportedSupportedΔ vs A: 0
ReasoningSupportedSupportedΔ vs A: 0
Model providerQwenQwen
Serving precision differs across offersSupportedSupportedΔ 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

Qwen3.5 397B A17B

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$0.39$2.34Unknown

Qwen3.8 27B

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$0.425$2.55$0.085
Migration preflight · compare each candidate with model A

Keeping Qwen3.5 397B A17B is a valid decision. No pairwise comparison certifies a drop-in replacement.

Qwen3.5 397B A17B → Qwen3.8 27B

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 Qwen3.5 397B A17B

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.

  • Qwen3.5 397B A17BEstimated total: $1.95 1,000 full attempts · base rates
  • Qwen3.8 27BEstimated total: $2.125 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 tokensQwen3.5 397B A17BQwen3.8 27BCoverage
2,000$1.95Lowest known$2.125All computed
4,096$2.76744Lowest known$3.0158All computed
32,768$13.94952Lowest known$15.2014All computed
131,072$52.28808Lowest known$56.9806All computed
Vary output; hold input, requests, cache and retries fixed.
Output tokensQwen3.5 397B A17BQwen3.8 27B
128$1.07952$1.1764
2,048$5.57232$6.0724
8,192$19.94928$21.7396

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 Qwen3.5 397B A17B a better-and-cheaper replacement for Qwen3.8 27B?

Qwen3.5 397B A17B has an equal-or-higher measured score and an equal-or-lower estimated price. Switching from Qwen3.8 27B loses capability or fails a requirement. Review the migration checks. The reported score intervals overlap; this is not an established quality improvement.

Which model is higher rated: Qwen3.5 397B A17B or Qwen3.8 27B?

Qwen3.5 397B A17B scores higher on LMArena general (1437.6 ±3.0) compared to Qwen3.8 27B (1437.4 ±4.4), a margin of 0.2 Elo points. However, because the 95% confidence intervals overlap, this does not establish an unconditional superiority across all tasks.

Which model is more cost-effective: Qwen3.5 397B A17B or Qwen3.8 27B?

On a standard balanced workload (3 prompt tokens per 1 generated token), Qwen3.5 397B A17B costs $0.878 per 1M tokens versus $0.956 for Qwen3.8 27B. Qwen3.8 27B is approximately 1.1× the price of Qwen3.5 397B A17B. Base input rates are $0.390 for Qwen3.5 397B A17B and $0.425 for Qwen3.8 27B; base output rates are $2.34 for Qwen3.5 397B A17B and $2.55 for Qwen3.8 27B.

How do their context windows and generation limits compare?

Qwen3.5 397B A17B provides a 262,144-token context window with a maximum output limit of 81,920 tokens. Qwen3.8 27B provides a 1,024,000-token context window with a maximum output of 131,072 tokens. Qwen3.8 27B offers a 3.9× larger context capacity, making it suitable for larger repositories and multi-document workflows.

What capabilities are lost when switching between Qwen3.5 397B A17B and Qwen3.8 27B?

Switching from Qwen3.5 397B A17B to Qwen3.8 27B causes no detected architectural capability loss. Conversely, moving from Qwen3.8 27B to Qwen3.5 397B A17B involves: Context window: 1,024,000 → 262,144 tokens; Output limit: 131,072 → 81,920 tokens. Run test suites on your specific prompt templates and tool declarations before cutting over production traffic.

Can Qwen3.5 397B A17B or Qwen3.8 27B be self-hosted or deployed privately?

Qwen3.8 27B provides open weights, enabling self-hosting on private GPU infrastructure, local quantization, and private data pipelines. Managed inference is also offered across third-party API providers (8 for Qwen3.5 397B A17B, 23 for Qwen3.8 27B).

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