Qwen·Qwen/Qwen3-4B-Instruct-2507
Qwen3 4B Instruct 2507 is ModelCap current-board rank #151 with Index 35.4 (Modeled · peer benchmarks), from the ModelCap Index as of 22 Sept 2026, 03:10 UTC.
Qwen3 4B Instruct 2507 has a published context limit of 262K tokens in the ModelCap catalogue as of 22 Sept 2026, 03:10 UTC.
Qwen3 4B Instruct 2507 weight access is open weights, from ModelCap classification of published repository metadata as of 22 Sept 2026, 03:10 UTC.
Snapshot facts · download the public dataset · methodology
OpenRouter endpoint status could not be refreshed. This model is not treated as currently served until a successful observation arrives.
Current provider status is unavailable.
Structured rows reported in this exact model repository at the pinned revision below. They are the publisher talking about itself: never measured evidence, never a score input on their own. Rows marked as used were laddered against independently measured peers inside the launch-card channel of this model's modeled placement, with the replay-measured optimism haircut applied; the others played no part.
No general-family board lists this model yet. Its point estimate is the precision-weighted mean of every channel that spoke for it; each channel’s share is the weight it carried. Independent measurements replace this placement as boards list the model.
BES normalizes admitted public benchmark evidence for measured Index rows. Its legacy score and observables prior do not define the public language rank.
These adoption and deployment observations remain context only. They do not change this model's ModelCap Index score or rank.
The bounded 0–100 capability score blends each source’s competitive placement with its published achievement, then combines capability families. Evidence breadth adds a modest uncertainty adjustment; price and popularity are not part of benchmark-led rank.
Experimental estimate. This layer predicts from admitted evidence and safe metadata only. It never enters ModelCap Score or rank.
Missing: artifactReproducibility.base-lineage, deployability.independent-provider-records, deployability.measured-provider-uptime, deployability.published-provider-pricing, evaluationProvenance.admitted-evaluation-observations, evaluationProvenance.capability-family-breadth, evaluationProvenance.evaluation-source-diversity, evaluationProvenance.identity-and-source-confidence, evaluationProvenance.evaluation-dataset-revisions.
A metadata completeness and deployability index—not a safety certification, quality grade, or production-readiness claim.
Market Gravity uses OpenRouter’s full-catalogue popularity order. Sparse Vercel top-ten observations appear only when published and do not affect the score.
This model is visible because it is available in the live catalogue, but it has no trustworthy community result yet. ModelCap does not infer a quality score from price, features, or market activity.
OpenRouter weekly popularity across the full model catalogue.
Independent providers versus the model's open or closed cohort, weighted by uptime.
Hugging Face 30-day downloads for open models; neutral for closed models.
Time since first public availability, on a six-month half-life.
Open weights. Downloadable weights are published under an OSI or free-culture license in the current repository metadata. This describes weight availability, not whether the full training stack qualifies as Open Source AI.
Models within 3 places of #151 on the ModelCap Index.
The same publisher's other current models with a public Index position.
Qwen3 4B Instruct 2507 holds ModelCap Index position #151 of 248 ranked language models as of 22 September 2026, with an Index score of 35.4 (interval 23.0–47.8); evidence: modeled · peer benchmarks. The Index combines public benchmark boards with published uncertainty; the methodology page explains the weighting.
No live API price is listed for Qwen3 4B Instruct 2507 in the current snapshot as of 22 September 2026 because the latest endpoint refresh was unavailable. ModelCap shows a price only when a provider currently lists one.
Qwen3 4B Instruct 2507 has a published context window of 262K tokens (262,144). Individual providers can serve less than the published maximum; the providers table lists each endpoint's own limit.
Open weights (apache-2.0). Downloadable weights are published under an OSI or free-culture license in the current repository metadata. This describes weight availability, not whether the full training stack qualifies as Open Source AI. The repository is Qwen/Qwen3-4B-Instruct-2507 on Hugging Face.
Qwen3 4B Instruct 2507 has published results on 1 public board tracked by ModelCap as of 22 September 2026: BFCL V4 35.7 (#40 of 83). Each board page ranks every tracked model on that benchmark; the ModelCap Index combines them with published uncertainty.
Qwen3 4B Instruct 2507 ranks #151 (Index 35.4) and granite 4.2 30b ranks #150 (Index 36.2) on the ModelCap Index as of 22 September 2026. The head-to-head page lines up their benchmarks, listed prices, context windows and provider counts side by side.
Qwen3 4B Instruct 2507 first appeared in the catalogue on 5 Aug 2025. It is the current version in its family. Rank and price movements since then are recorded on the site's changes feed.