Releasesby Ibisarize-phoenix-sqlean: v0.2.0
UpdatetodayUnverifiedAdded Oct 13, 2026
0.2.0 (2026-10-12) Features phoenix-sqlean: bundle SQLite 3.54.0 and support Python 3.15 (fff25bc)
More phoenix releases
Every phoenix releasearize-phoenix: v20.21.0
20.21.0 (2026-10-11) Features assign a random gradient to projects created by ingestion (8946f7a) cost-tracking: support Mistral AI model pricing (0e33736), closes #16878 Bug Fixes agents: upgrade pydantic-ai for CVE-2026-107286 (3cdb004) deps: update arize-phoenix-evals to 3.9.2 (e8aa1ea) playground: sync built-in model prices and model lists (35f86e5)...
arize-phoenix: v20.20.0
20.20.0 (2026-10-08) Features harbor: prompt hill-climb task on a text-to-SQL dataset (bd78550) mcp/sql: allowlist the annotation, evaluator, prompt, split and job tables (a37b841) mcp: add GraphQL schema, query, and mutation tools (9351580) playground: add claude-haiku-5-5 and remove retired Anthropic models (b978e7c) pxi: make PXI's MCP code mode...
arize-phoenix: v20.19.0
20.19.0 (2026-10-01) Features datagen: send each recorded application to its own project (d99a4db) tracing: add the DECISION span kind across the platform (43d1a54) Bug Fixes honor external resources flag for WASM and UI (7a71a21) Documentation remove legacy evaluation cookbooks (4506880)
arize-phoenix: v20.18.0
20.18.0 (2026-09-30) Features tracing: show eval results in the trace tree (2516fdb)
arize-phoenix: v20.17.0
20.17.0 (2026-09-30) Features agents: mark bash tool spans as errors on non-zero exit codes (cd0512e) graphql: add sort and sequenceNumbers to Dataset.experiments (9c14c5a) rest: add sequence_number, sort_dir, and sequence_numbers to list_experiments (305e1a9) Bug Fixes agents: leave the user out of the internal GraphQL scope when no one is logged in...
arize-phoenix: v20.16.0
20.16.0 (2026-09-23) Features add gpt-6-sol, gpt-6-luna, and claude-opus-5-5 to the cost tables and the playground (59d09f4) js: add dataset split write helpers (aab229f) js: add secrets management helper (c282563) spans: sort the span list endpoint by start_time (03f62e1) Bug Fixes agents: close out the PXI turn trace when a server tool raises (3a3d57b)...
arize-phoenix: v20.15.0
20.15.0 (2026-09-21) Features agents: mount configured skills in MCP and PXI (0809595) agents: search the GraphQL schema from phoenix-gql (2f58f8d) Bug Fixes cost: update built-in model token prices (d5535d4) db: preserve SQLite autoincrement counters across migrations (d3d7ce0) Documentation Add Phoenix release notes , 2026-09-16 (c9a6367) add redirects...
arize-phoenix-evals: v3.9.0
3.9.0 (2026-09-21) Features evals: add retrieval relevance evaluator (7efa7b2) evals: add session-level pii_detection evaluator (0e3b2c2) evals: completeness evaluator (63805d0) evals: deprecate document relevance evaluators (d67ea3f) Bug Fixes evals: accept role aliases in PromptTemplate message lists (767847c) evals: use non-blocking sleep in async rate...
Also shipped on Oct 13, 2026
Arize AI in October 2026What's included:
- Three training methods: Supervised fine-tuning (SFT) , direct preference optimization (DPO) , and reinforcement learning with group relative policy optimization (GRPO) . Each method trains a low-rank adapter (LoRA) or the full model weights. Custom loss functions : Train with any objective you compute in PyTorch from the model's log-probabilities.
LangSmith SDK 0.14.8
- fix(py): send session_id on RunTree.patch fix: keep punctuation out of freeze guard links chore(deps): bump source-map-js from 1.2.1 to 1.2.2 in /js fix(py): make the /runs/multipart 404 fallback to /runs/batch reachable feat(sandbox): accept any JOSE asymmetric algorithm chosen by the JWKS key (python) feat(sandbox): customer-managed KMS keys for S3...
Retry jobs with narrower API token permissions
Retry automation no longer needs the broader write_builds REST API scope. The new retry_builds scope allows a token to retry failed jobs in existing builds or individual eligible jobs, without granting access to the create-build or rebuild endpoints. When creating or editing an API access token, select Build Retries ( retry_builds ).
Post-training
Post-training lets you write your own training loop and run it against Together AI GPUs. Your code builds each batch and chooses the loss function and learning rate, while the GPUs you provision run the forward and backward passes, optimizer steps, and sampling. It bills per GPU-minute.
2026-10-13 updates
See what changed in Sourcegraph.
truss 0.18.34
- Add include and exclude filters to BDN mount configuration feat: Forward egress_restrictions for BIS and warn about missing hosts Release 0.18.34