Record supports_tool_search on the Bedrock Claude entries in both cost
map files and have _supports_tool_search_on_bedrock read it first via
the provider-resolved capability lookup, keeping the name patterns as a
fallback for ARNs and ids the map cannot resolve. Threads the flag
through ModelInfoBase and drops a dated remark from the pattern list
- formatInstallCommand now produces /plugin install {name}@litellm instead of /plugin marketplace add {source}
- extraKnownMarketplaces snippet fixed: source must be a nested object not a flat string; the flat string caused Claude Code to reject the settings file
- marketplace key renamed from my-org to litellm to match the name the proxy returns in marketplace.json
- setup tab now shows /plugin marketplace add command as primary option with settings.json as secondary
- usage tab now shows a hint to run /plugin marketplace update litellm when a plugin is not found
The /model/new endpoint returns a 400 validation error (type: validation_error)
when 'rpm and tpm must be set to a positive value when enforce_rpm_tpm_on_model_add
is enabled in general_settings' but the frontend's titleFor() keyword matcher
mistitled it as 'Rate Limit Exceeded' because the message contains 'rpm'/'tpm'
substrings, which matched the generic rate-limit keyword check before the more
specific validation check could catch it.
Add "'enforce_rpm_tpm_on_model_add' is enabled" to VALIDATION_MATCH so this
message is classified as a Validation Error, matching the actual HTTP 400
validation_error the backend already returns. A narrow match on the setting
name (rather than the generic "must be set when") avoids overriding the
status-based classification of unrelated 401s, e.g. the PKCE
'GENERIC_CLIENT_ID must be set when PKCE is enabled' error.
Add general_settings toggle 'enforce_rpm_tpm_on_model_add' (default false).
When true, /model/new rejects a model whose rpm or tpm is missing or not a
positive value, so the Admin UI Add Model form surfaces a 400 validation
error instead of silently storing an unbounded model (or one with a
zero/negative limit that would exclude it from routing).
Fixes#36487
video_edit, video_extension, and video_remix called request.body() after
user_api_key_auth had already parsed multipart/form bodies via
_read_request_body(), causing RuntimeError Stream consumed and 500s for
OpenAI SDK clients.
Use _read_request_body consistently and normalize bare-string or JSON-string
video references from form fields into video_id.
Claude Code drives Opus 4.7 with thinking {"type": "adaptive"} plus
output_config {"effort": "max"}. The anthropic-to-openai adapter
forwarded thinking verbatim for Claude models but dropped output_config,
and Bedrock Converse streams zero reasoningContent blocks for adaptive
thinking without an effort tier. Forward the effort subset of
output_config for Bedrock targets, accept it in the converse supported
params, and map it with the model's effort ceiling applied. Re-enable
the skipped e2e compat cell that catches this
Bedrock InvokeModel rejects tool_search_tool_* tool types unless the
request body carries the tool-search-tool-2025-10-19 beta. The model
allowlist gating that beta omitted Haiku 4.5 (and Opus 4.7, supported
since launch per live verification), so every tool-search request on
those models got a Bedrock 400. Add both to the allowlist and re-enable
the e2e compat cell that caught it.
* fix: net prompt-caching savings against the cache-write premium
Prompt-caching savings priced only the cache-read discount and ignored what
the provider charges to create the cache entry. Anthropic bills cache writes
at 1.25x the input rate, so a request that writes a large cache and reads
little from it is a net loss that the dashboard reported as a gain -- or, on
a pure cold write, as a flat zero.
The counterfactual the number answers is "what would this have cost with
caching off", where every token is billed at the input rate. Since
prompt_tokens partitions disjointly into text + reads + writes, that gives
savings = reads * (input - read_rate) - writes * (write_rate - input)
The write term is the premium over the input rate, not the full write cost:
the tokens would have been paid for at the input rate anyway, so only the
markup is attributable to caching.
The premium stays signed rather than clamped. Three models in the pricing map
price writes below input, and clamping would silently drop that saving.
A model with no cache_creation_input_token_cost falls open to the input cost,
yielding a zero premium -- this is why the change is a no-op for the implicit
caching providers (OpenAI, Gemini), which publish no write price, and bites
exactly on Anthropic and Bedrock.
Verified live through the proxy on a mock Anthropic rig across four cases
(cold pure-write, warm pure-read, write-heavy, read-heavy). Reported total
matched the derived net to the cent, including the negatives; the read-only
case is unchanged.
Pre-existing rows are not backfilled, so a range spanning the deploy mixes
gross and net.
* fix: read a zero cache-write price as unpublished, not free
deepseek-chat carries a literal 0.0 cache_creation_input_token_cost. The
fall-open only caught None, so the zero was taken at face value and the
premium became 0 - input_cost -- reporting a fabricated saving of
writes * input_cost on traffic that cached nothing.
No provider gives cache writes away, so a falsy price means the same thing
an absent one does.
* test: pin that the read leg keeps a literal zero price
The two zero prices mean opposite things and the asymmetry was unpinned.
A free cache write is unpublished pricing; a free cache read is real, and
15 models charge for input while serving reads for nothing. Copying the
write leg's falsy fall-open onto the read leg would zero out their savings.
* refactor: resolve caching rates through the established pricing helpers
Addresses Greptile's P1 and P2, and replaces hand-rolled pricing lookup with
the patterns this file and the cost calculator already own:
- Deployment pricing first: rates now resolve through _effective_model_info
(Router.get_deployment_model_info), the same helper the autorouter driver
uses, falling back to _model_info public rates. A deployment with negotiated
cache rates previously priced at the public map -- a 3x error on the repro.
- Individual prices read via _get_cost_per_unit, the cost calculator's
accessor, which also coerces string prices from config.yaml and resolves
service-tier suffixes; the previous raw .get() handled neither.
- Pricing tests no longer monkeypatch litellm.get_model_info; each case now
pins a real pricing-map entry with a fixture-drift assertion, and the
deployment-rate case follows the existing Router-fixture test pattern.
Behaviour on public rates is unchanged: 101 tests pass, including the exact
same live-verified formula.
* fix(cost-optimization): computeCacheLeakage divides net savings by all cached tokens, not reads alone
prompt_caching_savings_spend is net of the cache-write premium since PR #36452.
computeCacheLeakage was still dividing by cache_read_tokens alone, which:
1. Overstates the per-token rate on traffic that writes and reads cache equally:
a 1:1 read:write key shows rate = 0.002, not 0.001, if net savings is /bin/zsh.002
2. Flips the sign on write-heavy traffic: when writes cost more than reads save
(common on Anthropic and Bedrock), the aggregate net can go negative, but
dividing by reads alone would show a positive 'potential savings' for keys
that don't cache yet — recommending they start caching when it's currently
losing money overall
Fix: divide realizedCachingSavings by (cacheReadTokens + cacheCreationTokens),
matching the semantic that a key starting to cache pays those write premiums too.
When the rate is non-positive, price nothing (potentialSavings stays null, renders
as '—'), reusing the existing no-data fallback path. The card can't meaningfully
estimate savings from a losing rate.
Rename discountPerToken → netSavingsPerCachedToken to surface the semantics and
prevent this drift in future.
Update Usage tab and Cache Leakage card tooltips to describe net-of-premium cost.
Add tests for 1:1 read:write traffic and write-heavy negative-net traffic.
A deployment that carries one provider's credentials while resolving to
another is silently broken: litellm ignores the credentials and sends the
request to the resolved provider. The common shape is a Bedrock model
group where one entry lost its route prefix, so `model: claude-sonnet-5`
with aws_region_name set resolves to the first-party Anthropic API and
returns "x-api-key header is required". Because the router load balances
across the group, only the fraction of requests routed to that entry
fails, which reads as an intermittent provider outage rather than a
config error, and nothing at startup says otherwise.
Warn at deployment registration when provider-scoped credential params
(aws_*, vertex_*) sit on a model that resolves elsewhere, naming the
params, the resolved provider, and the likely missing prefix. Warn only:
an operator may be overriding a route deliberately, so this must not
block startup. Deployments litellm cannot classify are left alone.
Resolves LIT-5391
The iterations branch in AnthropicConfig.calculate_usage summed
cache_creation_input_tokens but never aggregated the per-iteration
cache_creation 5m/1h breakdown, leaving cache_creation_token_details
as None. As a result all cache-creation tokens fell back to the flat
5m write rate, underbilling 1h cache writes by up to 2x.
Aggregate the ephemeral_5m/ephemeral_1h split across iterations so 1h
writes are priced at the 1h rate.
Fixes LIT-4868
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
The mateo-berri PAT now arrives via systemd LoadCredential as a file
instead of the EnvironmentFile, so pytest, the proxy, and the
model-driven claude CLI never inherit it and a same-UID /proc read
cannot lift it. run_daily.sh reads the credential when present, still
accepts an exported GITHUB_TOKEN for manual runs, and dies up front
when publishing is enabled with neither. Full CLI sandboxing is
tracked in LIT-5420
The existing web_search cells drive Claude Code's client-side WebSearch
tool, which the CLI executes itself and feeds back as a tool_result. The
CLI never emits a web_search_20250305 definition, so those cells stayed
green while the Anthropic-managed server tool 400'd on Bedrock.
Add a cell that posts the server tool to a Bedrock deployment over
/v1/messages and asserts a web_search_tool_result block comes back, and
reword the compat row so it no longer reads as coverage of the server
tool. Model the server tool as a composed base shared with tool_search.
Resolves LIT-5391
The web-search interception hooks stamp
_websearch_interception_emit_native_blocks and
_websearch_interception_converted_stream onto kwargs to carry state across
the agentic loop, but neither was registered in all_litellm_params. The
param builder sweeps anything it does not recognize into the outbound
request, so a provider that validates its body rejects the whole call:
Bedrock Converse answers "_websearch_interception_emit_native_blocks:
Extra inputs are not permitted" with a 400, which breaks every request
interception touches on that route.
Register both alongside their code-interpreter counterparts, which were
already listed for exactly this reason.
Resolves LIT-5391
