stealthmodels.

MODEL BENCHMARKS

Ling 3.1 Flash

MIT-licensed hybrid reasoning MoE with 560B total and 25B active parameters, released FP8 weights, coding and tool-use support.

BENCHMARK BY BENCHMARK

Ling 3.1 Flash benchmark scores

StealthMark
42.4/100Overall rank 37/58
General Intelligence
35.6/100
Visual Intelligence
51.5/100

Ling 3.1 FlashOther modelsMedian

Reasoning efficiency

How many reasoning tokens a model needs for the quality it delivers.

Single-step reasoningHow much it reasons to understand a taskRank 35/56
Multi-step reasoningHow much it keeps reasoning as a task goes onRank 38/54

Cost efficiency

How much Performance you get for your money.

Rank 5/52

DATASHEET

Ling 3.1 Flash specs and pricing

A good fit forCoding, multi-step analysis and tool-using agents through hosted gateways or a high-memory multi-GPU deployment.

Ling 3.1 FlashOther modelsMedian

PriceInput / output per 1M tokens
$0.30 / $0.90Cheaper than 36 of 46 othersFrom $0.075 / $0.225 with other providers
Context windowTokens in one request
262,144Larger than 1 of 49 others
Output speedTokens per second, our measurement
115 tokens/sFaster than 20 of 51 others
Developer
InclusionAI
Released
29 Sep 2026
Family
Ling 3.1
License
MIT
Open model
Yes
Local model
No
Origin
China
Parameters
560B
Active parameters
25B
Input → output
Text → text
Architecture
Hybrid-linear MoE; 72 layers (63 KDA, 9 gated MLA); 512 routed experts, 8 selected and 1 shared.

Self-host Ling 3.1 Flash

Hardware
Multi-GPU server

Official FP8 weights on Hugging Face

All prices and providers2

Per 1M tokens: OpenRouter / Novita $0.075 input, $0.225 output and $0.015 cached input at current discounted rates; Vercel $0.30 input, $0.90 output and $0.06 cached input. Observed 2026-10-16.

  • Vercel AI GatewayObserved 2026-10-23
    Input $0.3Output $0.9Cached read $0.06

    per 1M tokens

    inclusionai/ling-3.1-flash
  • OpenRouterObserved 2026-10-23
    Input $0.075Output $0.225Cached read $0.015

    per 1M tokens

    inclusionai/ling-3.1-flash
Deployment details

Multi-GPU server; documented SGLang deployments on 8 H200 or 4 GB300 GPUs.

The developer advertises 1M context; the released configuration and current gateways specify 262,144 tokens. OpenRouter lists up to 235,929 output tokens; Vercel caps output at 32,768. Self-hosting requires memory for the weights, runtime and the selected workload.

Measurements and sources11

GAME BENCHMARKS

Watch Ling 3.1 Flash play

Foresight

Anticipating threats and planning ahead

43.1/100Rank 49/54
1 of 3 played
Foresight: complete playing field
Level 26 ptsTurn 14 / 18

Points per level

  1. Level 1Second ThoughtsPending
  2. Level 2Pressure Pot6 ptsBest 13Out of lives
  3. Level 3Hide or FlightNot played

Collect coins while avoiding ghosts. Ghosts move before the player each turn, and collisions cost a life. 3 levels in sequence, up to 66 turns in total. Earlier moves and outcomes carry forward, testing learning across levels.

Latch

Cooperative planning and learning from experience

45.9/100Rank 43/55
3 of 5 played
Latch: complete playing field
Level 49.8 ptsTurn 21 / 40

Points per level

  1. Level 1Hold the door5.2 ptsBest 5.2Cleared
  2. Level 2Crossfeed9.4 ptsBest 10.2Cleared
  3. Level 3Relay vaultPending
  4. Level 4Passing bay9.8 ptsBest 18.2Finished early
  5. Level 5Second thoughtsNot played

Coordinate two robots to push numbered cores onto matching sockets. Pressure plates hold doors open; moving either robot uses one turn. 5 levels in sequence, up to 152 turns in total. Earlier moves and outcomes carry forward, testing learning across levels.

Ling 3.1 Flash benchmark results ↗

StealthMark score

StealthMark measures how well a model’s abilities carry over to original, demanding tasks it could not have trained for: physical and spatial reasoning, planning across many steps, and precise visual work. How StealthMark works explains every benchmark, what it tests and why it is hard.

General Intelligence

The ability to reason, plan, solve problems and understand how things relate and behave. A high score means it thinks hard problems through, from careful analysis to plans that span many steps.

Visual Intelligence

The ability to understand visual intent and turn ideas, descriptions and plans into coherent, effective visuals. A high score means design ideas come out the way you meant them, in illustrations, icons, diagrams or interface visuals.

Overall score

Overall shows how well a model’s abilities carry over to original, demanding tasks, weighting every benchmark into one score. Estimated means the total includes an estimate for one missing scene.

Judging

Scores reflect how well each result meets the benchmark requirements. Game scores come from the recorded play: the points earned on each level and how each level ended. How StealthMark keeps scores honest.

Games

Each game is a sequence of levels, and earlier moves carry into later levels. A level that was not played is shown as not played and never counts as zero. Play every level of Latch or Foresight yourself to get a score calculated like the models’.

Reasoning efficiency

Good results with little thinking score high. On each task, a model’s result and reasoning tokens are compared with every other model and combined with the Balanced Integration Score (Liesefeld & Janczyk, 2019); 50 is average. Single step shows how much reasoning a model needs to understand and solve one problem, multi-step how efficiently it keeps working on a task by building on its earlier reasoning, as in agentic workflows. Not part of StealthMark.

Cost efficiency

How much StealthMark a model delivers for its money, in StealthMark points per dollar: its overall score divided by what one benchmark task costs it on average. The cost is what our runs were charged, or their tokens at the model’s list price; benchmarks a model has not run yet are estimated from its other costs. Like StealthMark, it is absolute and compares with no other model. Not part of StealthMark.

BENCHMARK RESULTS

Compare results

This link opens the same results in the same order.

StealthMark benchmark results

Reset benchmark preferences?

Return to the default view, clear comparison selections and restore the advertisement.