OpenAI compatible API · Attested · Public status

Weights & Biases

Explore Weights & Biases models on TrustedRouter with current routes, token pricing, policy sources, privacy posture, regional availability, and API support.

Verify gateway

Onebase URL to migrate

100sof models and routes

0prompt or output logs. Always.

Weights & Biaseswandb

No provider claim

All providers

These privacy labels describe Weights & Biases, the upstream model provider. Confidential requires all three: verified provider compute, end-to-end encryption and explicit zero data retention. Account and billing metadata are outside the content-retention claim.

Provider	Weights & Biases
API operator country	US Legal home of the API operator, not an inference-location guarantee.
Headquarters	Not verified
Routing status	Active
Provider website	https://wandb.ai/site/inference/
Models	18 public models
Credits routes	18
Zero data retention	no
Verified confidential inference	Not verified
Policy note	W&B Serverless Inference runs on CoreWeave infrastructure. W&B does not publish a zero-retention, confidential-compute, or end-to-end-encryption commitment for this service, so these routes use the Standard privacy tier. TrustedRouter does not enable W&B Weave tracing for provider calls. Policy source

Inference locations

Provider API declarations are also available on each model's endpoint API. Direct chat and Responses replies separate advertised availability from serving location in usage.inference_location; the actual serving region remains unknown when not reported.

Where the upstream GPUs execute the model, not the provider's headquarters, API ingress, or data storage. ZDR and confidential inference do not by themselves restrict geography.

Declared countries or regions: Not verified
No inference-location declaration has been verified for this provider.
Can locations change?: Not verified. Do not assume that the serving location is fixed.
Provider region pinning: Not verified.
Pinning through TrustedRouter: No verified provider-specific location pin documented here.
Infrastructure declaration: Inference location not verified; no country-specific residency commitment.
Evidence: Not yet reviewed.

Provider selection alone is not a region pin. TrustedRouter's US provider filter selects company jurisdiction, not GPU location. A regional gateway hostname does not pin a downstream GPU. For a binding residency requirement, confirm the model, region, failover scope and enforced controls before sending sensitive data.

Measured performance

27 samples

Continuously sampled across Weights & Biases's routed models: p50 TTFT, effective throughput, and success rate. Effective throughput uses provider-reported output tokens over complete request time. Unsupported route and probe-configuration rows are separated from provider downtime. No prompt or output content stored.

p50 TTFT	1382 ms
Effective throughput	80 tok/s n=8
Uptime	100.00%

Model	p50 TTFT	Effective throughput	Uptime	Config excluded	Availability samples
deepseek/deepseek-v4.1-flash	836 ms	—	100.00%	—	2
openai/gpt-oss-20b	925 ms	—	100.00%	—	3
qwen/qwen3.6-35b-a3b	1051 ms	—	100.00%	—	1
meta-llama/llama-3.3-70b-instruct	1181 ms	—	100.00%	—	4
z-ai/glm-5.3-flash	1241 ms	—	100.00%	—	1
ibm-granite/granite-4.2-8b	1253 ms	—	100.00%	—	2
deepseek/deepseek-v4-flash-0731	1382 ms	—	100.00%	—	3
google/gemma-4-26b-a4b-it	1421 ms	—	100.00%	—	1
google/gemma-4-31b-it	1741 ms	64 tok/s n=2	100.00%	—	3
openai/gpt-oss-120b	2037 ms	—	100.00%	—	1
minimax/minimax-m3	2054 ms	146 tok/s n=1	100.00%	—	3
meta-llama/llama-3.1-8b-instruct	2223 ms	—	100.00%	—	1
qwen/qwen3.8-27b	2834 ms	—	100.00%	—	1
z-ai/glm-5.2	2878 ms	203 tok/s n=1	100.00%	—	1
moonshotai/kimi-k2.6	—	71 tok/s n=2	—	—	0
moonshotai/kimi-k2.7-code	—	90 tok/s n=2	—	—	0

Weights & Biases performance history · Full provider & model leaderboard.

Provider models

Models served by Weights & Biases.

Each row links to pricing, provider, benchmark, and API pages for the model.

Model	AI IQ	Context	Input	Cached input	Output
`deepseek/deepseek-v4-flash-0731` DeepSeek: DeepSeek V4 Flash 0731	—	1,048,576	$0.13715/1M	$0.07385/1M	$0.2954/1M
`deepseek/deepseek-v4.1-flash` DeepSeek: DeepSeek V4.1 Flash	IQ 126#21	1,048,576	$0.211/1M	$0.03165/1M	$0.68575/1M
`google/gemma-4-26b-a4b-it` Google: Gemma 4 26B A4B	IQ 100#113	262,144	$0.1055/1M	$0.05275/1M	$0.3165/1M
`google/gemma-4-31b-it` Google: Gemma 4 31B	IQ 101#110	262,144	$0.1055/1M	Not published	$0.3587/1M
`ibm-granite/granite-4.2-8b` IBM: Granite 4.2 8B	IQ 93#144	131,072	$0.1055/1M	$0.05275/1M	$0.15825/1M
`meta-llama/llama-3.1-8b-instruct` Meta: Llama 3.1 8B Instruct	—	131,072	$0.2321/1M	Not published	$0.2321/1M
`meta-llama/llama-3.3-70b-instruct` Meta: Llama 3.3 70B Instruct benchmarks	—	131,072	$0.74905/1M	Not published	$0.74905/1M
`minimax/minimax-m3` MiniMax: MiniMax M3 benchmarks	IQ 115#54	524,288	$0.24265/1M	$0.05275/1M	$1.0128/1M
`moonshotai/kimi-k2.6` MoonshotAI: Kimi K2.6 benchmarks	IQ 115#52	262,144	$0.68575/1M	$0.15825/1M	$3.59755/1M
`moonshotai/kimi-k2.7-code` MoonshotAI: Kimi K2.7 Code benchmarks	IQ 112#66	262,144	$0.74905/1M	$0.15825/1M	$3.6925/1M
`nvidia/nemotron-3-ultra-550b-a55b` NVIDIA: Nemotron 3 Ultra	—	262,144	$0.5275/1M	$0.1055/1M	$2.26825/1M
`nvidia/nemotron-3.5-lightning-30b-a3b` nvidia/Nemotron-3.5-Lightning-30B-A3B	—	262,144	$0.07385/1M	$0.0422/1M	$0.211/1M
`openai/gpt-oss-120b` OpenAI: gpt-oss-120b	IQ 98#126	131,072	$0.03165/1M	Not published	$0.17935/1M
`openai/gpt-oss-20b` OpenAI: gpt-oss-20b	IQ 94#140	131,072	$0.03165/1M	Not published	$0.13715/1M
`qwen/qwen3.6-35b-a3b` Qwen: Qwen3.6 35B A3B	IQ 104#100	262,144	$0.26375/1M	Not published	$1.31875/1M
`qwen/qwen3.8-27b` Qwen: Qwen3.8 27B	IQ 116#50	65,536	$0.422/1M	$0.15825/1M	$3.165/1M
`z-ai/glm-5.2` Z.ai: GLM 5.2 benchmarks	IQ 119#45	1,048,576	$0.8018/1M	$0.1477/1M	$2.5531/1M
`z-ai/glm-5.3-flash` Z.ai: GLM 5.3 Flash	IQ 120#42	1,048,576	$0.15825/1M	$0.05275/1M	$0.5275/1M

Questions

Does Weights & Biases have zero data retention?

TrustedRouter does not currently mark Weights & Biases as provider-level zero data retention. Use trustedrouter/zdr or provider.min_privacy=zdr to select a different eligible route, and review the linked policy source for changes.

Is Weights & Biases end-to-end encrypted?

TrustedRouter does not currently mark Weights & Biases as end-to-end encrypted at the provider boundary. The TrustedRouter gateway is still attested, but the selected provider normally receives the request in order to run the model. Use trustedrouter/e2e for the stronger route requirement.

Which Weights & Biases models are available through TrustedRouter?

This page currently lists 18 public Weights & Biases models, with live pricing, route count, context length, measured performance when available, and links to each model's provider and benchmark pages.