Qwen3.8 Max

Qwen3.8 Max is developed by Alibaba. MeetNewAI lists 7 benchmark boards and 3 plans for it.

Plans and real price

Cheapest measured$0.040per million tokens, measuredDroid Max · medium confidence · Sep 29, 2026
PlanMonthly priceUsable tokens / monthReal price / MTokConfidenceData date
Droid Max$2005.1B$0.040mediumSep 29, 2026
Command Code GOAT$1062M$0.161mediumSep 30, 2026
OpenCode Go$1046.5M$0.215mediumSep 30, 2026

Data: real-api-pricingSources and licenses

Benchmarks

Artificial Analysis Intelligence Index v4.3 Intelligence Index · Oct 1, 2026

ConfigurationHarnessScoreCost per task
Qwen3.8 Max—40.2—

Terminal-Bench 4.0 (AA) Resolution Rate % · Oct 1, 2026

ConfigurationHarnessScoreCost per task
Artificial Analysis - Qwen3.8 MaxArtificial Analysis18.7—

Code Arena | WebDev Overall Arena Score · Sep 30, 2026

ConfigurationHarnessScoreCost per task
qwen3.8-max—1671.0 (1659.0–1683.0)—

Agent Arena · Overall (Ranking / Pareto) Net Improvement % · Sep 30, 2026

ConfigurationHarnessScoreCost per task
Qwen3.8 Max—2.8 (2.1–3.4)—

Artificial Analysis Coding Agent Index v1.5 Coding Agent Index · Oct 1, 2026

ConfigurationHarnessScoreCost per task
Claude Code - Qwen3.8 MaxClaude Code43.3$3.48

OpenDesign Arena · Quality ranking Average task score · Oct 1, 2026

ConfigurationHarnessScoreCost per task
Qwen 3.8-MaxOpenDesign72.0$0.595

DeepSWE v1.1 Pass@1 % · Sep 3, 2026

ConfigurationHarnessScoreCost per task
mini-swe-agent - qwen3-8-max (xhigh)mini-swe-agent57.5 (54.8–60.1)$3.73

Data: real-api-pricingSources and licenses