Qwen3.8 2.4T A95B

Qwen3.8 2.4T A95B is an open-weight sparse mixture-of-experts model from Qwen and the open-weight variant of Qwen3.8 Max, with 95 billion active parameters out of 2.4 trillion total. It is suited for coding, research, complex reasoning, and agentic workflows.

alibaba/qwen3.8-2.4t-a95b

Organization: Alibaba
Family: qwen
Providers: 25
Context: 262,144
Output limit: 131,072
Knowledge: —
Release: 2026-08-12
Updated: —
Weights: Open
Input: text
Output: text
Capabilities: —

Providers

Compare every cataloged offer for this model, including its provider, route, token limits, price, and capabilities.

33 offers · USD per million tokens

Provider	Model ID	Context	Output	Input / output · 1M	Reasoning	Tools	Structured	Details
AIHubMix	qwen3.8-2.4t-a95b	262,000	131,072	— / —	—	—	—	View
AI/ML API	alibaba/qwen3.8-2.4t-a95b	262,144	52,429	— / —	—	—	—	View
Alibaba	qwen3.8-2.4t-a95b	262,144	131,072	— / —	—	—	—	View
Alibaba (China)	qwen3.8-2.4t-a95b	262,144	131,072	— / —	—	—	—	View
Charm Hyper	qwen3.8-2.4t-a95b	1,000,000	128,000	— / —	—	—	—	View
Cortecs	qwen3.8-2.4t-a95b	262,144	262,144	— / —	—	—	—	View
Deep Infra	Qwen/Qwen3.8-2.4T-A95B	262,144	131,072	— / —	—	—	—	View
DevPass (LLM Gateway)	qwen3.8-2.4t-a95b	1,010,000	131,072	— / —	—	—	—	View
Eden AI	qwen/qwen3.8-2.4t-a95b	1,000,000	131,072	— / —	—	—	—	View
Fireworks AI	accounts/fireworks/models/qwen3p8-2p4t-a95b	262,144	131,072	— / —	—	—	—	View
Hugging Face	Qwen/Qwen3.8-2.4T-A95B	262,144	131,072	— / —	—	—	—	View
IteraCompute	qwen/qwen3.8-2.4t-a95b	1,048,576	131,072	— / —	—	—	—	View
Kilo Gateway	qwen/qwen3.8-2.4t-a95b	1,010,000	909,000	— / —	—	—	—	View
LLM Gateway	together-ai/qwen3.8-2.4t-a95b	1,010,000	909,000	— / —	—	—	—	View
LLM Gateway	novita/qwen3.8-2.4t-a95b	1,000,000	131,072	— / —	—	—	—	View
Merge Gateway	qwen/qwen3.8-2.4t-a95b	262,144	1,010,000	— / —	—	—	—	View
Modal	Qwen/Qwen3.8-2.4T-A95B	1,010,000	131,072	— / —	—	—	—	View
Novita AI	qwen/qwen3.8-2.4t-a95b	1,000,000	131,072	— / —	—	—	—	View
Alibabavia OpenRouter	qwen/qwen3.8-2.4t-a95b unknown	1,000,000	131,072	$2.00 / $6.00	Yes	Yes	Yes	View
DeepInfravia OpenRouter	qwen/qwen3.8-2.4t-a95b fp4	262,144	131,072	$2.00 / $6.00	Yes	Yes	Yes	View
Modalvia OpenRouter	qwen/qwen3.8-2.4t-a95b unknown	1,000,000	262,144	$2.00 / $6.00	Yes	Yes	Yes	View
Novitavia OpenRouter	qwen/qwen3.8-2.4t-a95b unknown	1,000,000	131,072	$2.00 / $6.00	Yes	Yes	Yes	View
SiliconFlowvia OpenRouter	qwen/qwen3.8-2.4t-a95b fp8	1,048,576	131,072	$2.00 / $6.00	Yes	Yes	Yes	View
Togethervia OpenRouter	qwen/qwen3.8-2.4t-a95b unknown	1,010,000	909,000	$2.00 / $6.00	Yes	Yes	Yes	View
Venicevia OpenRouter	qwen/qwen3.8-2.4t-a95b unknown	262,144	65,536	$2.00 / $6.00	Yes	Yes	No	View
Opper	qwen3.8-2.4t-a95b	262,144	131,072	— / —	—	—	—	View
Requesty	qwen3.8-2.4T-A95B@eu	1,000,000	262,144	— / —	—	—	—	View
Requesty	qwen3.8-2.4T-A95B	262,144	262,144	— / —	—	—	—	View
RunInfra	Inferact/Qwen3.8-2.4T-A95B-NVFP4	262,144	32,768	— / —	—	—	—	View
SiliconFlow	Qwen/Qwen3.8-2.4T-A95B	1,049,000	131,000	— / —	—	—	—	View
TensorX	qwen/qwen3.8-2.4t-a95b	262,144	64,000	— / —	—	—	—	View
Venice AI	qwen-3-8-2-4t-a95b	262,144	65,536	— / —	—	—	—	View
Vercel AI Gateway	alibaba/qwen3.8-2.4t-a95b	262,144	128,000	— / —	—	—	—	View

Prices and limits apply to each offer. A dash means the value is not provided. OpenRouter offers show the hosting provider and routing channel separately.

Pricing

Compare token rates and cache charges across available routes and hosting configurations.

USD per million tokens

Input from$2.00Per 1M input tokens

Output from$6.00Per 1M output tokens

Available offers33Across 25 providers

Lowest listed input and output prices across the offers above; they may belong to different providers. These are base rates, before conditional pricing, discounts or additional fees.

Endpoint price comparisonLatest observed snapshot · 7 lowest-priced configurations

InputOutput

Alibabaalibaba

$2.00$6.00

DeepInfradeepinfra/fp4

$2.00$6.00

Modalmodal

$2.00$6.00

Novitanovita

$2.00$6.00

SiliconFlowsiliconflow/fp8

$2.00$6.00

Togethertogether

$2.00$6.00

Venicevenice

$2.00$6.00

qwen/qwen3.8-2.4t-a95bObserved Oct 10, 2026, 11:31 AM UTC

Input: $2.00
Output: $6.00
Cache read: $0.25

Hosting prices through OpenRouter All endpoints and cache rates

Endpoint	Input / 1M	Output / 1M	Cache read / 1M	Cache write / 1M
Alibabaalibaba	$2.00	$6.00	$0.25	$2.50
DeepInfradeepinfra/fp4	$2.00	$6.00	$0.20	—
Modalmodal	$2.00	$6.00	$0.25	—
Novitanovita	$2.00	$6.00	$0.25	—
SiliconFlowsiliconflow/fp8	$2.00	$6.00	$0.25	—
Togethertogether	$2.00	$6.00	$0.25	—
Venicevenice	$2.00	$6.00	$0.25	—

Each row is a hosting configuration. Open an offer in Providers for its full pricing conditions and additional fees. Endpoint observations may differ from the routing catalog's base price.

Performance

Review reported latency and output-throughput percentiles from the latest observed rolling window.

Latest 30-minute observation

Time to first token · milliseconds

Latency distributionRolling 30-minute snapshot · 6 endpoints

P50P75P90P99

Alibabaalibaba

1,806–5,337.26 ms

DeepInfradeepinfra/fp4

926–3,278.26 ms

Modalmodal

384–1,567.3 ms

Novitanovita

1,571–15,043.74 ms

SiliconFlowsiliconflow/fp8

2,061.5–31,644.93 ms

Togethertogether

962.5–13,050.63 ms

All latency percentiles 7 endpoints

Endpoint	P50	P75	P90	P99
Alibabaalibaba	1,806 ms	2,183 ms	2,744.2 ms	5,337.26 ms
DeepInfradeepinfra/fp4	926 ms	1,166.25 ms	1,611 ms	3,278.26 ms
Modalmodal	384 ms	487 ms	848.6 ms	1,567.3 ms
Novitanovita	1,571 ms	1,924.5 ms	2,499.6 ms	15,043.74 ms
SiliconFlowsiliconflow/fp8	2,061.5 ms	2,948 ms	5,340.7 ms	31,644.93 ms
Togethertogether	962.5 ms	1,861.75 ms	3,681.7 ms	13,050.63 ms
Venicevenice	54,752 ms	88,578 ms	107,554.2 ms	119,369.52 ms

Output throughput · tokens per second

Throughput distributionRolling 30-minute snapshot · 6 endpoints

P50P75P90P99

Alibabaalibaba

29–46.99 t/s

DeepInfradeepinfra/fp4

30–122 t/s

Modalmodal

84–227 t/s

Novitanovita

31–48.13 t/s

SiliconFlowsiliconflow/fp8

40–51 t/s

Togethertogether

42–230.64 t/s

All throughput percentiles 7 endpoints

Endpoint	P50	P75	P90	P99
Alibabaalibaba	29 t/s	34 t/s	39 t/s	46.99 t/s
DeepInfradeepinfra/fp4	30 t/s	45 t/s	85 t/s	122 t/s
Modalmodal	84 t/s	108 t/s	152 t/s	227 t/s
Novitanovita	31 t/s	36 t/s	42 t/s	48.13 t/s
SiliconFlowsiliconflow/fp8	40 t/s	46 t/s	47 t/s	51 t/s
Togethertogether	42 t/s	67 t/s	111 t/s	230.64 t/s
Venicevenice	49 t/s	95 t/s	119.4 t/s	143.88 t/s

Percentiles describe the observed request distribution. These are snapshots of a rolling window, not a historical time series.

Uptime

Compare successful request rates across the latest 5-minute, 30-minute, and 24-hour observations.

Successful requests by window

Availability by endpointCurrent overlapping windows · 7 endpoints

Endpoint5 minutes30 minutes24 hours

AlibabaalibabaObserved Oct 10, 2026, 11:31 AM UTC

100%

DeepInfradeepinfra/fp4Observed Oct 10, 2026, 11:31 AM UTC

100%

ModalmodalObserved Oct 10, 2026, 11:31 AM UTC

100%

99.72%

NovitanovitaObserved Oct 10, 2026, 11:31 AM UTC

100%

SiliconFlowsiliconflow/fp8Observed Oct 10, 2026, 11:31 AM UTC

—

100%

TogethertogetherObserved Oct 10, 2026, 11:31 AM UTC

100%

VeniceveniceObserved Oct 10, 2026, 11:31 AM UTC

—

100%

94.88%

Reported by OpenRouter at the time of collection; rate-limited requests are excluded. A dash means no measurement was reported. This is not a live availability check.

Benchmarks

Review attributed evaluation results on each source's original scale.

External evaluations

qwen/qwen3.8-2.4t-a95bObserved Oct 10, 2026, 11:31 AM UTC

Benchmark profileArtificial Analysis · shared source scale from 0 to 80

Coding

71.9

Agentic

50.1

Intelligence

39.9

Sources: Artificial Analysis and Design Arena, via OpenRouter. Scores retain their original scales and are not OpenAI Suite ratings.

Detailed evaluations

Evaluation scoresAccuracy or item F1 · higher is better

gpqa diamondQwen: Qwen3.8 2.4T A95B

87.079%

tau bench verified airlineQwen: Qwen3.8 2.4T A95B

75.619%

Evaluation	Configuration	Source	Results	Details
Composite indices	Qwen3.8 2.4T A95Bqwen/qwen3.8-2.4t-a95b-20260812	Artificial Analysis	Intelligence 39.9 · Coding 71.9 · Agentic 50.1	Details
gpqa diamond	Qwen: Qwen3.8 2.4T A95Bqwen/qwen3.8-2.4t-a95b-20260812	OpenRouter	87.079%	Details
tau bench verified airline	Qwen: Qwen3.8 2.4T A95Bqwen/qwen3.8-2.4t-a95b-20260812	OpenRouter	75.619%	Details

Sources: Artificial Analysis, Design Arena and OpenRouter. Evaluation configurations and scales differ; results are not interchangeable. Updated Oct 24, 2026.

Technical details & API

Explore specifications, supported parameters, reasoning controls, and API identifiers for each OpenRouter route.

Current route specifications

qwen/qwen3.8-2.4t-a95bObserved Oct 10, 2026, 11:31 AM UTC

Context: 1,048,576
Maximum output: 131,072
Tokenizer: Qwen
Input: text
Output: text
Moderation: No
Added to OpenRouter: 2026-08-12

Base URLhttps://openrouter.ai/api/v1

Model IDqwen/qwen3.8-2.4t-a95b

Version IDqwen/qwen3.8-2.4t-a95b-20260812

Weights IDQwen/Qwen3.8-2.4T-A95B

Reasoning

Required: Yes
Enabled by default: Yes
Default effort: xhigh
Supported efforts: xhigh, medium, low

Supported parameters

frequency_penalty
include_reasoning
logit_bias
logprobs
max_tokens
min_p
presence_penalty
reasoning
reasoning_effort
repetition_penalty
response_format
seed
stop
structured_outputs
temperature
tool_choice
tools
top_k
top_logprobs
top_p

Default parameters

top k: 20
top p: 0.95
temperature: 1

Frequently asked questions

Answers based on the model metadata and serving offers currently in the catalog.

How much does Qwen3.8 2.4T A95B cost?

Listed rates start at $2.00 per million input tokens and $6.00 per million output tokens. Prices vary by provider and configuration; see Pricing for details.

Which providers offer Qwen3.8 2.4T A95B?

There are 33 cataloged offers across 25 providers. Compare model IDs, prices and limits in Providers.

What is the context window?

The cataloged model context is 262,144 tokens. Each serving endpoint may apply a different limit.

Does it support tools and structured output?

Tool calling: Not reported. Structured output: Not reported. Support can vary by endpoint.

Resources

qwen3.8-max

Hugging Face ↗

Catalog history · 3 recorded revisions

Metadata observations since this model was first cataloged. These are not model release versions.

Oct 20, 2026 · 6f3f4bab9f4d
Sep 14, 2026 · df74b59d0f47
Sep 12, 2026 · 45457691124c