Codex custom Responses provider: can it point at Sume?
Codex CLI 0.162.0 adds custom Responses-compatible provider config. Sume's Agent Completions is not a model endpoint; use Sume's hosted MCP from Codex instead.
No. Codex CLI 0.162.0 (Oct 8) adds configuration for a custom Responses-compatible model provider, but Sume does not document a Responses-compatible endpoint. Agent Completions at POST /v1/agent/completions returns a 202 receipt for a run, with model fixed to sume-agent. To use Sume from Codex, add the hosted MCP server.
What did Codex add?
The Codex changelog lists 0.162.0 with managed Git worktree tools, task pinning and custom Responses-compatible provider configuration. 0.161.0 (Oct 7) added a /mcp login <name> terminal command and enterprise MCP sign-in.
| Version | Date | Relevant items |
|---|---|---|
| 0.162.0 | Oct 8 2026 | Custom Responses-compatible provider config; managed worktree tools; task pinning |
| 0.161.0 | Oct 7 2026 | /mcp login <name>; enterprise MCP sign-in |
Why is Agent Completions not a provider?
A model provider answers a prompt with tokens. Agent Completions starts a Sume agent run:
- It returns 202 with an
agent.runreceipt and astatus_urlto poll. modelaccepts onlysume-agent.- It takes exactly one of
instructionormessages[], and rejects assistant turns. generation_spend_cap_usdis required.- Service-account keys are refused.
What should Codex use?
Codex as the agent, Sume as the tools. Add https://mcp.sume.com/mcp as a remote server, sign in with /mcp login or use a key, and let Codex call generate_video, jobs_wait and the rest. Choose Agent Completions when you want Sume to host the run and call you back with a signed agent.run.terminal event. Details are in Agent Completions.
Sources
Related posts
More in Comparisons
- Context stitching and streaming TTS: Eleven v4 vs Sume async jobs
Eleven v4 lists context stitching and bidirectional streaming. Sume TTS is async and non-streaming, so keep a section in one job and join sections.
- Deepgram Flux Multilingual $0.0078/min: 7,500 minutes vs Sume STT $75
Deepgram lists Flux Multilingual streaming at $0.0078 a minute. For 7,500 minutes that is $58.50, against $75.00 on Sume STT. The full Deepgram ladder is below.
- ElevenLabs Scribe v2 $0.22/hour vs Sume STT $0.60/hour: 7.5 hours
Scribe v2 lists $0.22 per hour and Sume STT is $0.01 per minute, or $0.60 per hour. For 7.5 hours of audio: $1.65 vs $4.50, 2.7x apart.
- Scribe v2 Realtime $0.39 vs batch $0.22: 37 hours vs Sume STT
ElevenLabs charges 77 percent more for Scribe v2 Realtime than Scribe v2. For 37 hours the gap is $14.43 against $8.14, and Sume STT is $22.20.
Written by Sume