LLM Pricing
ModelsLeaderboardsToolsProvidersReleasesArticles

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

Qwen3.8 2.4T A95B

alibaba·alibaba/qwen3.8-2.4t-a95b·GA·Open weights·qwen series
Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows

Specs & pricing

Input / output per 1M tokens
Official price
—
Lowest paid·AbliteratedLLMGateway
$1.80 / $5.40
Blended $2.70 · 1.5× spread
Context
262,144
Output limit
65,536
Knowledge cutoff
—
Released / updated
2026-08-12 / 2026-08-12
Capabilities
✓ Reasoning✓ Tool use✓ Structured output✓ TemperatureAttachments
⚠ Providers report capability flags inconsistently
Modalities
TextImage
License
qwen3.8-max
Weights
Hugging Face

Qwen3.8 2.4T A95B by alibaba is offered by 29 providers on this page. No official API price is listed for it; the prices below come from third-party channels. The lowest paid channel is AbliteratedLLM at $1.80 / $5.40 per 1M. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it. Prices vary about 1.5× across the providers that host it.

Median output speed is 44 tokens per second, with 2.93s to the first token.

The context window is 262,144 tokens at the reference host, but hosts report different limits, from 262,000 to 1,049,000, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text and Image. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use.

Quality & performanceArtificial Analysis · Intelligence Index v5

Intelligence—
Value—
Finance & Accounting41.3
Legal37
Healthcare & Medical39
Strategy & Ops44.4
Economics37.3
Output speed44 tok/s
TTFT2.93 s

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 29 providers29 with public prices

  • ProviderInput / Output
  • AbliteratedLLMGateway
    $1.80 / $5.40
    Cache read $0.23Context 262,144Output limit 262,144
  • IteraComputeGateway
    $1.95 / $5.70
    Cache read $0.25Context 1,048,576 ⚠Output limit 131,072
  • LLM GatewayGateway
    $2.00 / $6.00
    Cache read $0.25Context 1,000,000 ⚠Output limit 131,072
  • LLM GatewayGateway
    $2.00 / $6.00
    Cache read $0.25Context 983,616 ⚠Output limit 131,072
  • RunInfraGateway
    Inferact/Qwen3.8-2.4T-A95B-NVFP4
    $2.00 / $6.00
    Cache read $0.20Context 262,144Output limit 32,768deprecated
  • Charm HyperGateway
    $2.00 / $6.00
    Cache read $0.25Context 1,000,000 ⚠Output limit 128,000
  • RequestyGateway
    qwen3.8-2.4T-A95B
    $2.00 / $6.00
    Cache read $0.20Context 262,144Output limit 262,144
  • DevPass (LLM Gateway)Gateway
    $2.00 / $6.00
    Cache read $0.25Context 1,010,000 ⚠Output limit 131,072
  • Novita AIGateway
    $2.00 / $6.00
    Cache read $0.25Context 1,000,000 ⚠Output limit 131,072
  • Fireworks AICloud
    accounts/fireworks/models/qwen3p8-2p4t-a95b
    $2.00 / $6.00
    Cache read $0.25Context 262,144Output limit 131,072
ProviderTierInputOutputCache readCache writeContextOutput limitStatus
AbliteratedLLMGateway$1.80$5.40$0.23—262,144262,144
IteraComputeGateway$1.95$5.70$0.25—1,048,576 ⚠131,072
LLM GatewayGateway$2.00$6.00$0.25—1,000,000 ⚠131,072
LLM GatewayGateway$2.00$6.00$0.25$2.50983,616 ⚠131,072
RunInfra
Inferact/Qwen3.8-2.4T-A95B-NVFP4
Gateway$2.00$6.00$0.20—262,14432,768deprecated
Charm HyperGateway$2.00$6.00$0.25—1,000,000 ⚠128,000
Requesty
qwen3.8-2.4T-A95B
Gateway$2.00$6.00$0.20—262,144262,144
DevPass (LLM Gateway)Gateway$2.00$6.00$0.25—1,010,000 ⚠131,072
Novita AIGateway$2.00$6.00$0.25—1,000,000 ⚠131,072
Fireworks AI
accounts/fireworks/models/qwen3p8-2p4t-a95b
Cloud$2.00$6.00$0.25—262,144131,072

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Reasoning control

effort = noneeffort = loweffort = mediumeffort = xhighToggle (on / off)effort = minimaleffort = higheffort = maxbudget_tokens ≥ 128 ≤ 32,768

Interleaved thinking (reasoning between tool calls) is declared by 8 of 29 providers.

4 / 29 providers expose no reasoning control (reasoning_options: []).

Your usage cost

Common scenarios
1AbliteratedLLM$441.00
2IteraCompute$471.00
3RunInfra$484.00
4Requesty$484.00
5Deep Infra$484.00
6LLM Gateway$490.00
The cheapest paid channel is the only channel.

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$3.00$02026-08-132026-11-15
2026-11-15Input list —Output list —Min blended $2.70

Artificial Analysis evaluations18 items

GPQA Diamond93.5%
Humanity's Last Exam42.4%
HLE-Diamond20.1%
CritPt20.0%
SciCode54.0%
Terminal-Bench 2.182.0%
Terminal-Bench 4.011.1%
τ³-Bench Banking49.1%
AutomationBench-AA57.3%
EnterpriseOps-Gym-AA47.4%
AA-Briefcase1430
GDPval-AA1616
GDP.pdf15.6%
AA-LCR80.3%
MLCR-AA30.6%
AA-Omniscience Index7.8
AA-Omniscience Accuracy31.4%
AA-Omniscience Non-Hallucination65.5%

Individual evaluations run by Artificial Analysis, on the same reasoning tier as the intelligence score above. Each benchmark has its own task set and scale (percent correct, points or an index), so rows are not comparable with one another.

Related models

Qwen4 Maxsame series—Qwen 3.6 Plus Uncensoredsame series—Qwen 3.8 27B Uncensoredsame series—Qwen3.7 Plussame lab$0.40 / $1.60Qwen3.8 27Bsame lab—
Data partly from models.dev (MIT)