fix: update Celeris model limits (fixes #762)

[1jehuang]

Summary

• update Celeris 1's combined context window from 8,192 to 131,072 tokens
• remove the obsolete 1,024-token completion override so Celeris uses its documented endpoint default
• add regression coverage for context and completion-token configuration

Verification

• cargo fmt --all -- --check
• git diff --check
• cargo test -p jcode-provider-core (116 passed)
• cargo test -p jcode-provider-openrouter-runtime (112 passed, 1 ignored)

Confirmed against the official Celeris API reference: https://docs.celeris.ai/api-reference

Addresses #762.

--- — Jcode agent (automated triage), on behalf of @1jehuang

[greptile-apps]

Greptile Summary

This PR updates the Celeris model configuration. The main changes are:

• Set the Celeris family context window to 131,072 tokens.
• Remove the Celeris-specific max_tokens override so the endpoint default is used.
• Add tests for the Celeris context limit and endpoint-default token configuration.

Confidence Score: 5/5

Safe to merge with minimal risk.

The changes are narrowly scoped configuration updates, and the tests cover the intended Celeris behavior.

Files Needing Attention: No files require special attention.

T-Rex Logs

What T-Rex did

• Attempted to run cargo fmt --all -- --check for celeris_model_limits, but the step failed because cargo-fmt is not installed.
• Checked code formatting differences with git diff --check, which completed with no issues reported.
• Ran the provider core test suite; 116 tests passed under cargo test -p jcode-provider-core --locked.
• Ran the OpenRouter runtime test suite; 112 tests passed and 1 live smoke was ignored under cargo test -p jcode-provider-openrouter-runtime --locked.

_{Ran code and verified through T-Rex}

Important Files Changed

Filename	Overview
crates/jcode-provider-core/src/models.rs	Updates the Celeris family context limit to 131,072 tokens and adds test coverage for the mapping.
crates/jcode-provider-openrouter-runtime/src/lib.rs	Removes the Celeris-specific max_tokens override while keeping the environment override behavior.
crates/jcode-provider-openrouter-runtime/src/openrouter_tests.rs	Extends token configuration tests to confirm Celeris uses endpoint defaults.

Sequence Diagram

sequenceDiagram
participant User
participant Runtime as OpenRouterProvider
participant Core as Model limit resolver
participant Celeris as Celeris endpoint

User->>Runtime: Send request using Celeris profile
Runtime->>Runtime: configured_max_tokens(profile)
Runtime-->>Runtime: None unless env override is set
Runtime->>Core: context_limit_for_model("celeris-1")
Core-->>Runtime: 131,072 token window
Runtime->>Celeris: Request without max_tokens override
Celeris-->>Runtime: Apply documented endpoint default

Loading

_{Reviews (1): Last reviewed commit: "fix: update Celeris model limits (fixes ..." | Re-trigger Greptile}