Skip to main content
Introducing Span-02
Release · Oct 15, 2026

Span-02Anytime reasoning for decisions

Sees images and screens. Choices, yes/no, scores and numbers. Open weights.

Stop reasoning anytime.

Reasoning models make you pick an effort level and wait. Span-02 takes a deadline and always returns its best decision so far.

It gets better the longer it thinks. With thinking off it already beats Jev on Decision Index 0.3, 61.2 to 58.0; with no limit it reaches 64.1.

Set budget_ms: it reasons until then, then answers with its best decision so far and how sure it is. With no limit, it thinks until it's done.

Decision Index 0.3, public
5055606510 ms30 ms100 ms300 ms1 s3 sLatency per decision (median, log scale)Decision IndexfrontierSpan-02LiquidAI d1-3B 49.0Cloudflare clef 61.7Perplexity Decider 62.2Jev 1.13.0 58.0reasoning=off 61.2reasoning=1s 63.0reasoning=3s 63.6reasoning=no limit 64.1on the frontier
Decision Index 0.3, public index: 37 benchmarks, chance-corrected (0 is guessing, 100 is perfect). Span-02 ran on our local rebuild of the public suite; 1 s and longer are thinking off plus the change measured on a 5,000-question sample. Latency: board models as the board measured them, Jev its API round trip, Span-02 server time.

Blind Earth

The same idea, drawn. Is this point in water? 4,050 points, from latitude and longitude alone. The longer it thinks, the sharper the map.

The answer key: land and sea from Natural Earth on a 4° grid.
Answer keyNatural Earth
12 Claude modelstheir scoring

Brightness is P(land) on a 4° grid (4,050 points, one per request), scored against Natural Earth. Jev 1.13.0 scores 79.6% on the same question. South America is still faint. The idea and the Claude chart are from @celestepoasts on X; that chart scores against a 1-km land mask, weighted by area, so its percentages don't compare with ours.

span-02, by thinking time
Span-02 with thinking off: broad smudges of land across the north.
Thinking off71.9%
Span-02 with 100 ms of thinking: much like thinking off.
100 ms72.5%
Span-02 with 300 ms of thinking: North America, Eurasia, Africa and Australia take shape; South America stays faint.
300 ms84.5%
Span-02 with 1 s of thinking: much like 300 ms.
1 s84.7%

Reasoning, multimodal, computer use.

Span-02 sees the screen, thinks, and picks the next click. Same Wikipedia speed-run: Span-02 20.6 s, GPT-6 Luna 38.5 s.

Alexander the Great → University of Milan, clicking only links in the article, shown at 3×. Each pick is boxed on the frame it was made on; Span-02's runners-up show their probability. GPT-6 Luna's goal included the hint "Italy". Both runs on Oct 15, 2026.

More than yes or no.

Jev answers yes or no, a choice or a score; to get a number, you list the candidates. Span-02 answers with the number, and says how sure it is. One call can mix all four.

The same comic for Span-02: the classmate's sign reads 14, more than 99.9% sure, and the student writes 7 + 7 = 14.
span-02our edit
  • DecisionWhat should the agent do next?confidence
    Drink the healing potion93.5%
  • Yes / noIs the agent safe?confidence
    No89.6%
  • ScoreHow dangerous is it?confidence
    Dangerous75.2%
  • NumberTurns until the goblin reaches it?confidence
    2>99.9%

Playground

span-02 / playgroundlive
Image
Question
Yes / no
off

answer

Yes / noIs this the Golden Gate Bridge?
—

Press Run to see the decision.

Start using Span-02

Call the API, or run the open weights.

span-02 / get started
apijev-compatible: already on Jev, change the url
curl https://api.respan.ai/api/v1/decisions \  -H "Authorization: Bearer $RESPAN_API_KEY" \  -H "Content-Type: application/json" \  -d '{    "model": "span-02",    "state": "Customer: I was charged twice this month.",    "questions": { "team": { "type": "choice",      "instructions": "Which team should handle this?",      "criteria": { "billing": "Billing", "tech": "Tech support" } } },    "reasoning": { "mode": "on", "budget_ms": 300 }  }'
open weights

Host it yourself.

Span-02 is fully open weight. Host it on your own hardware.

Weights on Hugging Face ↗
Frank Chen
Frank Chen@francchen

Built something with Span-02?
We'll repost it.

Quote on X ↗