Telnyx
Explore Telnyx models on TrustedRouter with current routes, token pricing, policy sources, privacy posture, regional availability, and API support.
Telnyxtelnyx
Telnyx Inference serves OpenAI-compatible chat completions on Telnyx-owned GPUs. Marketplace routes include Telnyx-hosted models only, not its third-party BYOK passthrough catalog.
These privacy labels describe Telnyx, 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 | Telnyx |
|---|---|
| API operator country | US Legal home of the API operator, not an inference-location guarantee. |
| Headquarters | Austin, Texas, United States Provider-submitted headquarters; company profile. Source; reviewed 2026-09-27. |
| Legal entity | Telnyx LLC; Illinois, USA limited liability company |
| Headquarters | Austin, Texas, USA |
| Registered and operating address | 600 Congress Ave, Floor 14, Austin, TX 78701, USA |
| DUNS | 966115342 |
| EIN | 27-0273220 |
| CEO | David Casem |
| API | OpenAI-compatible chat completions, streaming, tools, structured output and reasoning content on supported models. |
| Catalog | Authenticated native OpenAI catalog; not yet a Provider Reliability Contract v2 declaration. |
| Pricing | Input, output and cached-input rates refresh hourly from Telnyx's authenticated APIs. Listed routes use default service-tier prices; priority and flex tiers are not enabled through TrustedRouter. |
| Policy precedence | For Personal Data, DPA 12.1 gives precedence to the SCCs and UK Addendum, then the DPA, then the AI Addendum, then the remaining Agreement. Scope and exceptions remain governed by those documents. |
| Policy reviewed | September 22, 2026. Published endpoint-specific ZDR and DPA v3.0; provider statements, not hardware-attested guarantees. |
| Routing status | Active |
| Provider website | https://telnyx.com/products/inference |
| Models | 17 public models |
| Credits routes | 17 |
| Zero data retention | yes All routes |
| Verified confidential inference | Not verified |
| Policy note | Telnyx's published policy states zero data retention for Telnyx-hosted /v2/ai/openai/chat/completions and /v2/ai/openai/embeddings: content is processed in memory and discarded after the response; request metadata is retained. TrustedRouter currently publishes hosted chat routes and adapts Responses to chat completions. Stateful Responses, AI Assistants with conversation history, third-party models and /v2/ai/embeddings are outside this ZDR scope. OpenAI-compatible embeddings may run in any region. DPA 3.1 restricts training of models available to others, with exceptions for written opt-in or data de-identified under 3.5. This is policy-backed ZDR, not confidential compute or verified end-to-end attestation. 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
- United States; Europe (Telnyx EU region; countries unspecified); Australia; United Arab Emirates
Provider-wide GPU region set. Model availability is narrower; see the catalog snapshot below. Germany in the storage-locality documentation does not establish Germany as the inference country. - Can locations change?
- Dynamic, latency-based routing by default. Regional ingress is a preference; capacity events and failover can move inference to another available region. Model deployments may change.
- Provider region pinning
- Telnyx documents region + mode: strict on a matching regional ingress domain for its own hosted models. Requests fail if the selected region cannot serve them. Preferred mode is not a residency guarantee.
- Pinning through TrustedRouter
- Not supported by the current TrustedRouter integration: it uses api.telnyx.com without Telnyx's strict region controls. Do not send Telnyx-specific region/mode fields as a TrustedRouter pin.
- Infrastructure declaration
- Multi-region, dynamically routed among the model's available Telnyx regions; no fixed inference-region commitment on this TrustedRouter route.
- Evidence
- Public documentation and authenticated model catalog. Reviewed .
Advertised regions by model (16)
Catalog snapshot: . Refreshed by the catalog pipeline. Availability can change; these are not per-request receipts or enforced residency guarantees. Only models with a region declaration are shown.
- deepseek/deepseek-v4.1-flash
- Europe (Telnyx EU region; countries unspecified); United States
- google/gemma-2b-it
- United States
- meta-llama/llama-3.1-70b-instruct
- United States
- meta-llama/llama-3.1-8b-instruct
- United States
- meta-llama/llama-3.3-70b-instruct
- Europe (Telnyx EU region; countries unspecified); United States
- minimax/minimax-m2.7
- United States
- minimax/minimax-m3
- United States
- moonshotai/kimi-k2.5
- Australia; United States
- moonshotai/kimi-k2.6
- United States
- moonshotai/kimi-k3
- United States
- qwen/qwen3-235b-a22b
- United States
- qwen/qwen3.8-27b
- United States
- z-ai/glm-5.1
- United States
- z-ai/glm-5.2
- United States
- z-ai/glm-5.3
- Australia; Europe (Telnyx EU region; countries unspecified); United States
- z-ai/glm-5.3-flash
- Australia; Europe (Telnyx EU region; countries unspecified); United Arab Emirates; United States
- Inference regions and strict pinning
- Processing versus storage
- Native model catalog (authentication required)
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
72 samplesContinuously sampled across Telnyx'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 | 1046 ms |
|---|---|
| Effective throughput | 161 tok/s n=7 |
| Uptime | 100.00% |
| Model | p50 TTFT | Effective throughput | Uptime | Config excluded | Availability samples |
|---|---|---|---|---|---|
| z-ai/glm-5.3-flash | 1046 ms | — | 100.00% | — | 39 |
| qwen/qwen3.8-27b | 597 ms | — | 100.00% | — | 2 |
| moonshotai/kimi-k3 | 601 ms | 91 tok/s n=1 | 100.00% | — | 1 |
| google/gemma-2b-it | 942 ms | — | 100.00% | — | 3 |
| minimax/minimax-m3 | 943 ms | 161 tok/s n=2 | 100.00% | — | 1 |
| z-ai/glm-5.3 | 950 ms | 195 tok/s n=1 | 100.00% | — | 2 |
| meta-llama/llama-3.1-70b-instruct | 1204 ms | — | 100.00% | — | 3 |
| moonshotai/kimi-k2.5 | 1215 ms | — | 100.00% | — | 6 |
| deepseek/deepseek-v4.1-flash | 1238 ms | — | 100.00% | — | 1 |
| qwen/qwen3-235b-a22b | 1294 ms | — | 100.00% | — | 3 |
| z-ai/glm-5.2 | 1379 ms | 237 tok/s n=1 | 100.00% | — | 3 |
| meta-llama/llama-3.1-8b-instruct | 1391 ms | — | 100.00% | — | 2 |
| minimax/minimax-m2.7 | 1399 ms | — | 100.00% | — | 1 |
| z-ai/glm-5.1 | 1564 ms | — | 100.00% | — | 3 |
| meta-llama/llama-3.3-70b-instruct | 2313 ms | — | 100.00% | — | 2 |
| moonshotai/kimi-k2.6 | — | 69 tok/s n=2 | — | — | 0 |
Telnyx performance history · Full provider & model leaderboard.
Models served by Telnyx.
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-0731DeepSeek: DeepSeek V4 Flash 0731 |
— | 1,048,576 | $0.13715/1M | $0.03165/1M | $0.2743/1M |
deepseek/deepseek-v4.1-flashDeepSeek: DeepSeek V4.1 Flash |
IQ 126#21 | 1,048,576 | $0.3165/1M | $0.01/1M | $1.266/1M |
google/gemma-2b-itgemma-2b-it |
— | 8,192 | $0.211/1M | $0.211/1M | $0.211/1M |
meta-llama/llama-3.1-70b-instructMeta: Llama 3.1 70B Instruct |
— | 131,072 | $0.633/1M | $0.633/1M | $0.633/1M |
meta-llama/llama-3.1-8b-instructMeta: Llama 3.1 8B Instruct |
— | 131,072 | $0.211/1M | $0.211/1M | $0.211/1M |
meta-llama/llama-3.3-70b-instructMeta: Llama 3.3 70B Instruct |
— | 131,072 | $0.633/1M | $0.633/1M | $0.633/1M |
minimax/minimax-m2.7MiniMax: MiniMax M2.7 |
IQ 105#92 | 196,608 | $0.22155/1M | $0.03165/1M | $1.266/1M |
minimax/minimax-m3MiniMax: MiniMax M3 |
IQ 115#54 | 524,288 | $0.28485/1M | $0.0844/1M | $1.1605/1M |
moonshotai/kimi-k2.5MoonshotAI: Kimi K2.5 |
IQ 110#74 | 262,144 | $0.633/1M | $0.422/1M | $3.165/1M |
moonshotai/kimi-k2.6MoonshotAI: Kimi K2.6 |
IQ 115#52 | 262,144 | $0.701575/1M | $0.0844/1M | $4.22/1M |
moonshotai/kimi-k3MoonshotAI: Kimi K3 |
IQ 127#20 | 1,048,576 | $2.8485/1M | $0.28485/1M | $14.2425/1M |
qwen/qwen3-235b-a22bQwen3-235B-A22B |
— | 32,768 | $0.633/1M | $0.422/1M | $2.11/1M |
qwen/qwen3.8-27bQwen: Qwen3.8 27B |
IQ 116#50 | 65,536 | $0.422/1M | $0.05275/1M | $3.165/1M |
z-ai/glm-5.1Z.ai: GLM 5.1 |
IQ 113#59 | 202,752 | $1.0339/1M | $0.13715/1M | $4.642/1M |
z-ai/glm-5.2Z.ai: GLM 5.2 |
IQ 119#45 | 1,048,576 | $1.055/1M | $0.211/1M | $4.22/1M |
z-ai/glm-5.3Z.ai: GLM 5.3 |
IQ 125#24 | 1,048,576 | $1.31875/1M | $0.2532/1M | $4.22/1M |
z-ai/glm-5.3-flashZ.ai: GLM 5.3 Flash |
IQ 120#42 | 1,048,576 | $0.142425/1M | $0.028485/1M | $0.47475/1M |
Questions
Does Telnyx have zero data retention?
TrustedRouter records Telnyx as supporting provider-level zero data retention based on the policy source linked on this page. This is a provider policy claim, separate from TrustedRouter's content-stateless real-time gateway and from end-to-end confidential compute.
Is Telnyx end-to-end encrypted?
TrustedRouter does not currently mark Telnyx 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 Telnyx models are available through TrustedRouter?
This page currently lists 17 public Telnyx models, with live pricing, route count, context length, measured performance when available, and links to each model's provider and benchmark pages.