Reliability methodology · public facts

Only measured availability is published.

Model.sale publishes a model as live only when a recent compatibility check, usage signal and active price agree. This page explains the checks so a developer or search system can interpret a status correctly.

Protocol checkA synthetic request validates JSON status, response shape, request ID and terminal usage. Streaming checks verify first bytes, event ordering and a terminal event.
PerformanceWe measure connect time, time to first byte, first token and total latency. Public pages show only rounded, non-customer aggregate facts when volume is sufficient.
Billing gateUsage must be present or covered by a documented deterministic policy. Price and margin are evaluated before publication; a health check alone cannot publish a model.
Current validation snapshot

19 live · 33 temporarily unavailable

Generated from the same public registry used by the API catalog. Last observed catalog check: 2026-10-11T03:34:24.563Z. This is a point-in-time compatibility snapshot, not a historical uptime percentage.

ModelStatusHealthSynthetic latencyChecked
gpt-5.5LIVEhealthy · 1001126 ms2026-10-11T01:34:36.426Z
gpt-5.6-lunaLIVEhealthy · 1002254 ms2026-10-11T01:34:46.127Z
gpt-5.6-solLIVEhealthy · 1002471 ms2026-10-11T01:34:48.475Z
gpt-5.6-terraLIVEhealthy · 1001699 ms2026-10-11T01:34:39.599Z
gpt-6-astraLIVEhealthy · 1002393 ms2026-10-11T01:34:32.472Z
gpt-6-lunaLIVEhealthy · 1005093 ms2026-10-11T01:35:07.415Z
gpt-6-solLIVEhealthy · 1001664 ms2026-10-11T01:35:03.971Z
gpt-6.1-solLIVEhealthy · 1001787 ms2026-10-11T01:35:11.992Z
claude-fable-5-1LIVEhealthy · 1003622 ms2026-10-11T01:35:20.016Z
claude-haiku-4-5-20251001LIVEhealthy · 100871 ms2026-10-11T01:34:26.663Z
claude-opus-4-6LIVEhealthy · 1002255 ms2026-10-11T01:35:24.030Z
claude-opus-4-7LIVEhealthy · 100945 ms2026-10-11T01:35:22.555Z
claude-opus-4-8LIVEhealthy · 100895 ms2026-10-11T01:35:19.491Z
claude-opus-5LIVEhealthy · 1001608 ms2026-10-11T01:35:17.705Z
claude-opus-5-5LIVEhealthy · 1005117 ms2026-10-11T01:35:14.474Z
claude-sonnet-5-5LIVEhealthy · 1001435 ms2026-10-11T01:35:25.722Z
glm-5.2LIVEhealthy · 1006492 ms2026-10-11T01:34:53.973Z
glm-5.3LIVEhealthy · 1001241 ms2026-10-11T01:34:53.038Z
glm-5.3-flashLIVEhealthy · 1001120 ms2026-10-11T01:34:55.344Z
gpt-5.4UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
gpt-5.4-20UNAVAILABLEunhealthy · 034 ms2026-10-02T09:53:56.463Z
gpt-5.4-miniUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
claude-haiku-4-5UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
claude-opus-4-5UNAVAILABLEunhealthy · 035 ms2026-10-02T09:53:51.813Z
claude-sonnet-4-5UNAVAILABLEunhealthy · 034 ms2026-10-02T09:53:53.468Z
claude-sonnet-4-6UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
deepseek-flashUNAVAILABLEunhealthy · 0110 ms2026-10-11T01:34:55.916Z
deepseek-v4-flashUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
deepseek-v4-proUNAVAILABLEunhealthy · 0118 ms2026-10-11T01:34:57.288Z
glm-5UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
glm-5.1UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
qwen-3.8UNAVAILABLEunhealthy · 035 ms2026-10-02T09:53:58.119Z
qwen3.5-plusUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
qwen3.6-plusUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
qwen3.7-maxUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
qwen3.7-plusUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
kimi-k2.5UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
kimi-k2.6UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
kimi-k3UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
mimo-v2-omniUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
mimo-v2-proUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
mimo-v2.5UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
mimo-v2.5-proUNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
minimax-m2.5UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
minimax-m2.7UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
minimax-m3UNAVAILABLEunhealthy · 0—2026-10-11T03:34:24.563Z
gpt-4o-transcribeUNAVAILABLEunhealthy · 09880 ms2026-10-10T21:42:15.944Z
gpt-image-1.5UNAVAILABLEunhealthy · 040 ms2026-10-10T21:41:47.458Z
gpt-image-2UNAVAILABLEunhealthy · 0288 ms2026-10-10T21:41:45.762Z
gpt-image-2.5UNAVAILABLEunhealthy · 0—pending
gpt-image-2.5-flareUNAVAILABLEunhealthy · 035 ms2026-10-10T21:41:43.803Z
gpt-image-2.5-sunburstUNAVAILABLEunhealthy · 044 ms2026-10-10T21:41:41.863Z

What the health status means

StatusScoreMeaning
Healthy95–100Recent successful checks with normal latency and stream completion.
Degraded85–94Serving may continue while latency, 429s or synthetic results need attention.
UnhealthyBelow 85Customer publication is blocked or the model is temporarily unavailable.

The score is a compact operational signal, not a quality benchmark. It combines success, latency, streaming completion, rate-limit responses and synthetic checks.

Publication and refresh schedule

  1. Lightweight probes run every five minutes for protocol and endpoint health.
  2. A round-robin model sweep runs every eight hours; each result has a visible checked timestamp.
  3. A model stays public only while its check is within the freshness window, price is active and margin is at least 50%.
  4. After a failure, the catalog marks the exact model unavailable; it does not reroute the request to another model.
  5. Recovery requires another successful check and an audited publication action.

Privacy of measurement

Probes use synthetic prompts and never copy customer requests into shadow traffic. We retain status, latency, token counts and safe error classes for operations. Prompts, source code and response bodies are not written to logs, traces, analytics or Sentry.

For current incidents and component availability, see status.model.sale. For price calculation, see the pricing methodology.