Model Radar

Every week our model scout indexes new LLMs on OpenRouter, cross-references public usage and benchmark coverage, then benches the candidates nobody has measured yet — prioritized by coverage gap. This is the latest scan.

Scanned 9/27/2026 · verdict: no upgrade recommended

Bench scorecard

Model	Score /10	Latency	Est. cost	Notes
inclusionai/ling-3.0-flash-sante:free	0	0.1s	free	Rate limit exceeded.

New free-tier models

nvidia/nemotron-3.5-lightning:free — usage-ranked
inclusionai/ling-3.0-flash-sante:free — no public benchmark
inclusionai/ling-3.0-flash-fin:free — usage-ranked
qwen/qwen3.8-27b:free — usage-ranked
liquid/lfm-2.5-2.6b:free — no public benchmark

New paid models worth watching

openai/gpt-6-luna-pro — usage-ranked
anthropic/claude-opus-5.5 — AA-indexed
x-ai/grok-4.7 — usage-ranked
xiaomi/mimo-v2.6-pro-ultraspeed — usage-ranked
fireworks/ember-1 — usage-ranked

Advisor notes

.

**Report:**
The evaluation suite failed to return usable data for the candidate `inclusionai/ling-3.0-flash-sante:free` due to systemic free-tier exhaustion (429 Rate Limit errors). 

**Fleet Status & Verdict:**
*   **Current Fleet Baseline:** `google/gemini-2.5-flash` and `gpt-oss-120b` remain our top performers with 100% success rates.
*   **Verdict:** No upgrades possible. As the candidate failed to benchmark, we maintain the current configuration.

**Benchmark Data:**
*   *Benchmark data: Artificial Analysis (artificialanalysis.ai)*
*   The candidate `inclusionai/ling-3.0-flash-sante` provided no benchmark score; therefore, it does not meet the "clear beat" threshold (>=1.5 points) required for an upgrade recommendation.

**Recommendation:**
Continue relying on `google/gemini-2.5-flash` for high-reliability tasks. Monitor the availability of `ling-3.0-flash-sante` if free-tier capacity returns, but avoid switching until a successful, high-scoring benchmark is recorded.

Benchmark data: Artificial Analysis (artificialanalysis.ai)

Machine-readable: GET /public/model-radar on the API returns this scan as JSON (CORS-open); ?scan=<scanId> reads an older one from GET /public/model-radar/scans.

The scout, bench, and advisor behind this page are LiveGraph graphs themselves — every score above is a real model call through the same dispatch path your own runs take. To see that engine in action, try the live demo — no signup, no key — then run the models above on your own graphs, on our keys or yours.