Item detail

github.com

BerriAI/litellm

BerriAI/litellm is a developer tool in RepoRadar's AI Gateway section, holding Gold tier and a 'try now' verdict. Its strongest signal is workflow potential, scored 10.0 out of 10.

Produced and maintained by RepoRadar Research · Methodology · Corrections.

Open on GitHub →Back to radar

Score9.0

Popularity97.0

Riskconditional

TierGold

Score breakdown

Usefulness9.0

Novelty8.0

Momentum9.0

Maturity9.0

Open-source/build8.4

Evidence7.2

Workflow potential10.0

Setup ease4.2

Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.

Why it matters

Useful if you need a single, auditable routing layer instead of wiring each model provider separately.

Where this stands now

BerriAI/litellm ranks #1 of 10 tracked AI Gateway items by composite score (9.0 against a section median of 7.9). The section currently carries 6 Gold, 4 Silver. RepoRadar has retained observations for this record since 2026-06-18 (91 days in the current window). Signal extremes versus the section: momentum at the 90th percentile; novelty at the 75th percentile.

Who should use it

builders shipping multi-vendor AI products platform teams managing provider costs and quotas ops teams standardizing LLM observability

Who should skip it

Skip BerriAI/litellm if the source link, documentation, or setup requirements do not align with your current workflow or stack.

About this signal

BerriAI/litellm is tracked by RepoRadar as a developer tool in the AI Gateway section. First seen 2026-06-18; the source record was last checked on 2026-06-18. The current verdict is 'try now' with a Gold tier and advanced setup difficulty. Across RepoRadar's eight signals, BerriAI/litellm is strongest on workflow potential (10.0) and practical usefulness (9.0) and weakest on setup ease (4.2) — a profile worth weighing against your own priorities. This page summarizes the public evidence on the linked source page and states where additional review is still needed.

How this item is evaluated

The BerriAI/litellm record combines a 9.0/10 composite score with separate popularity (97.0), risk (conditional), and setup (advanced) signals. See the scoring methodology for the current weights and evidence definitions.

Questions worth asking before you adopt this

Putting this into practice? Read How to evaluate an AI tool before you adopt it for the checklist behind this score.

Risk explanation

Gateway errors can become system-level failure points; protect provider credentials and test failover and allowlisting before production; The repository includes mixed licensing paths (MIT for core, check enterprise paths before bundling).

Evidence links

github.com

Closest alternatives / related signals

llm gateway multi-provider observability cost-control

Verification record

What RepoRadar actually verified

Tested in a bounded workflow Source-only update since test

Bounded representative workflow retained by RepoRadar verification harness. Last checked 2026-07-14T00:53:12.671649Z.

A source release dated 2026-09-16 is newer than the retained test dated 2026-07-14; the previous test may no longer represent the current release.

Read as a sequence: 2 retained check(s) between 2026-07-13 and 2026-07-14 with outcomes of 1 failed, 1 passed; evidence scopes covered bounded representative workflows and bounded setup checks. Recorded setup time totals 1 minute(s).

failed · cohort-20260712-litellm-import

Tester: RepoRadar automated local verification harness
Started: 2026-07-13T03:22:09.504383Z
Completed: 2026-07-13T03:22:34.676310Z
Environment: Windows 10 AMD64; Python 3.11.9; credential-stripped child environment; disposable home/cache
Install/setup time: 0 minute(s)
Evidence scope: Bounded setup or capability check
Cleanup: Per-check temporary home and work directory removed. Shared cohort package cache removed.

Actions exercised

Created a disposable home, work directory, and isolated package cache with credential-like environment variables excluded.
Executed bounded check: Install and import LiteLLM in an isolated Python environment.
Captured the complete sanitized stdout, stderr, exit status, and 25.17-second wall time.

Observed results

Check exited 1 after 25.17 seconds; failure output is retained and the verification stage must not advance.

Observed strengths

No product strength was established in this failed attempt; diagnostic evidence is retained.

Friction

The command emitted stderr; warnings or errors are preserved in the retained log for review.

Limitations

No proxy was started and no provider credentials or model requests were used.
This bounded cohort check is not a production benchmark or a claim of real user-workflow adoption.

Pricing assessment: No paid plan or metered provider usage was exercised; package or licensing, hosting, and provider costs remain workflow-dependent.

Privacy assessment: No repository content, user data, or provider prompt was transmitted; broader product data handling was not assessed by this bounded run.

Open retained test log →

passed · cohort-20260713-litellm-mocked-completion-workflow

Tester: RepoRadar automated local verification harness
Started: 2026-07-14T00:53:03.005597Z
Completed: 2026-07-14T00:53:12.671649Z
Environment: Windows 10 AMD64; Python 3.11.9; credential-stripped child environment; disposable home/cache
Install/setup time: 1 minute(s)
Evidence scope: Bounded representative workflow
Cleanup: Per-check temporary home and work directory removed. Shared cohort package cache removed.

Actions exercised

Created a disposable home, work directory, and isolated package cache with credential-like environment variables excluded.
Created 1 synthetic fixture file(s) inside the disposable work directory; retained hashes prove the exact inputs.
Constructed a system/user incident message pair and invoked LiteLLM 1.91.3's completion API with the explicit mock_response argument.
Validated LiteLLM's normalized model, assistant content, and finish-reason fields and retained a compact response artifact.
Executed bounded check: Send a two-message incident request through LiteLLM's completion API with its explicit local mock_response path and validate the normalized response object.
Captured the complete sanitized stdout, stderr, exit status, artifact checks, and 9.67-second wall time.

Observed results

Command exited 0 after 9.67 seconds.
LiteLLM returned a normalized ModelResponse whose assistant content, model name, stop reason, and two-message input count matched the fixture.
Expected marker 'CHECK_OK model=gpt-4o-mini queue=platform-oncall finish=stop messages=2' was observed in retained output.
Validated result.json: 4 required marker(s) present and 0 excluded marker(s) absent; size and SHA-256 are retained.

Observed strengths

The provider-neutral completion surface and normalized response object can be exercised deterministically without credentials when the explicit mock path is used.

Friction

The latest observed 1.92.0 source distribution attempted a Rust build and failed because link.exe was unavailable; this passing run is intentionally pinned to the prior pure-Python 1.91.3 release.
Setup or runtime emitted 5 stderr line(s); the complete warnings/errors are preserved in the retained log.

Limitations

The pinned 1.91.3 mock-response path does not contact, authenticate to, route among, retry, stream from, or measure any model provider. LiteLLM 1.92.0 was separately probed on this Windows Python 3.11 environment and could not build because the source distribution required an unavailable MSVC linker, so this run does not establish latest-release compatibility.
This credential-free disposable workflow does not establish operator use, production scale, model quality, reliability under sustained use, or team adoption.

Pricing assessment: The completion used LiteLLM's explicit mock_response path and no model endpoint, token billing, proxy, or hosted service.

Privacy assessment: The two synthetic messages stayed inside LiteLLM's local mock path; no API key existed and no provider request was made.

Open retained test log →

Verification sources

Official source ↗ · checked 2026-09-17

Longitudinal intelligence

How this decision record is moving

Raw history JSON →

71 dated snapshots retained from 2026-06-18 through 2026-09-17; see the snapshot index for explicit coverage gaps. Stars, version, release, pricing, integration, risk, maintenance, verdict, score, and momentum fields remain explicit even when a source has not reported them. Repository momentum is a normalized 0–10 RepoRadar signal; GitHub stars appear only where the popularity monitor retained exact timestamped observations.

RepoRadar score9.0 current · +0.0 net

Repository momentum9.0 current · +0.0 net

GitHub stars (observed)59,008 current · +5,488 net

GitHub stars59,008 exact observation

Versionv1.103.0-dev.1

Last release2026-09-16T02:56:03Z

Maintenanceactive

Current riskconditional

Current verdicttry now

Pricing baselineNo structured commercial pricing baseline

Pricing checkedNot applicable or not recorded

Pricing freshnessNo dated commercial pricing review

Integrations baselineAnthropic, Azure OpenAI, LangChain, OpenAI

Recent dated points

Date	Score	Momentum	Stars	Risk	Verdict	Maintenance
2026-09-17	9.0	9.0	59,008	conditional	try now	active
2026-09-15	9.0	9.6	58,779	conditional	try now	active
2026-09-13	9.0	9.0	58,600	conditional	try now	active
2026-09-12	9.0	9.0	58,570	conditional	try now	active
2026-09-11	9.0	9.6	58,502	conditional	try now	active
2026-09-08	9.0	9.0	58,252	conditional	try now	active
2026-09-07	9.0	9.0	Not recorded	conditional	try now	not recorded
2026-09-05	9.0	9.6	58,043	conditional	try now	active
2026-09-04	9.0	9.6	58,043	conditional	try now	active
2026-09-03	9.0	9.0	Not recorded	conditional	try now	not recorded
2026-09-02	9.0	9.0	57,766	conditional	try now	active
2026-09-01	9.0	9.0	57,766	conditional	try now	active

Why the record changed

stars changed2026-09-17

Stars changed: 58975 → 59008.

version changed2026-09-17

Version changed: v1.102.0-rc.1 → v1.103.0-dev.1.

stars changed2026-09-17

Stars changed: 58779 → 58975.

stars changed2026-09-15

Stars changed: 58600 → 58779.

version changed2026-09-13

Version changed: v1.102.0-dev.2 → v1.102.0-rc.1.

stars changed2026-09-13

Stars changed: 58570 → 58600.

stars changed2026-09-12

Stars changed: 58555 → 58570.

stars changed2026-09-12

Stars changed: 58502 → 58555.

stars changed2026-09-11

Stars changed: 58252 → 58502.

version changed2026-09-11

Version changed: v1.100.0 → v1.102.0-dev.2.

stars changed2026-09-08

Stars changed: 58245 → 58252.

stars changed2026-09-01

Stars changed: 57753 → 57766.