---
type: "post"
slug: "jalapeno-first-inference-results"
title: "Jalapeno first inference results"
description: "OpenAI published measured results for Jalapeño, its first custom inference chip, claiming 1.5-1.9x more work per watt and 1.7-3.6x lower latency on a public third-party benchmark."
retrieval_nugget: "OpenAI published measured results for Jalapeño, its first custom inference chip, claiming 1.5-1.9x more work per watt and 1.7-3.6x lower latency on a public third-party benchmark."
published_at: "2026-08-31"
updated_at: "2026-09-12"
record_date: "2026-08-25"
date_kind: "published_at"
topics: ["agent-retrieval","agents","ai","search"]
entities: ["OpenAI"]
source_url: "https://openai.com/index/jalapeno-first-results"
source_title: "Jalapeño’s first results show industry-leading speed and efficiency in AI inference"
source_domain: "openai.com"
source_terms: ["Jalapeno","first","inference","Jalape","performance"]
summary_word_count: 194
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:jalapeno-first-inference-results"
status: "published"
source_urls: ["https://openai.com/index/jalapeno-first-results"]
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:0fc7ec266c89241bf219ed2bab43b11aac7bd427571837de72a4e4138a51a0e1","reviewed_at":"2026-09-12T10:00:00Z","source_evidence_count":1,"verified_claim_count":2,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":2,"implication_count":1,"watch_condition_count":1,"related_record_count":3}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"OpenAI published first results for Jalapeño, its custom inference chip, on 25 August.","observed_facts":[{"text":"Jalapeño delivered 1.5 to 1.9 times more AI work per watt at peak throughput and 1.7 to 3.6 times lower end-to-end latency.","source_urls":["https://openai.com/index/jalapeno-first-results"]},{"text":"Testing covered GPT-OSS 120B, DeepSeek R1 and Kimi K2.5 1T on the InferenceX benchmark from SemiAnalysis.","source_urls":["https://openai.com/index/jalapeno-first-results"]}],"mechanism":"OpenAI evaluates at a matched user experience: how much useful work a system completes per unit of power while still meeting the latency interactive agents require.","why_now":"The company's stated reason is that agents run many steps in sequence, so per-step delay compounds across a task, which makes latency a throughput problem rather than a comfort metric.","implications":["The measurement method is the part that matters for agents."],"evidence_boundary":"The numbers are OpenAI's own runs against unnamed comparison systems, and ramp plans are stated as intent.","watch_conditions":["The condition to watch is whether independent InferenceX submissions reproduce the per-watt range, because a first-party benchmark result on a public benchmark is still a first-party result."],"related_records":[{"url":"https://newruntime.com/posts/gimlet-multi-silicon-inference-cloud","relation":"Prior coverage of inference as a multi-silicon scheduling problem."},{"url":"https://newruntime.com/posts/llm-inference-gpu-capacity-planning","relation":"Prior coverage of inference sizing as capacity planning."},{"url":"https://newruntime.com/posts/anthropic-model-hardware-standard-control-boundary","relation":"Prior coverage of hardware as a control boundary."}]}
routes: {"html":"https://newruntime.com/posts/jalapeno-first-inference-results/","markdown":"https://newruntime.com/posts/jalapeno-first-inference-results.md","json":"https://newruntime.com/posts/jalapeno-first-inference-results.json"}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
---

# Jalapeno first inference results

## Retrieval answer

OpenAI published measured results for Jalapeño, its first custom inference chip, claiming 1.5-1.9x more work per watt and 1.7-3.6x lower latency on a public third-party benchmark.

OpenAI published first results for Jalapeño, its custom inference chip, on 25 August. Measured on InferenceX, a public benchmark from SemiAnalysis, against leading commercially available systems, Jalapeño delivered 1.5 to 1.9 times more AI work per watt at peak throughput and 1.7 to 3.6 times lower end-to-end latency, rising to 2.1 to 4.1 times higher performance on highly interactive workloads. The tests ran across GPT-OSS 120B, DeepSeek R1 and Kimi K2.5 1T, so the architecture is not tuned to one in-house model.

The measurement method is the part that matters for agents. OpenAI evaluates at a matched user experience: how much useful work a system completes per unit of power while still meeting the latency interactive agents require. The company's stated reason is that agents run many steps in sequence, so per-step delay compounds across a task, which makes latency a throughput problem rather than a comfort metric.

That is the scheduling reality described in [Gimlet treating inference as a multi-silicon problem](https://newruntime.com/posts/gimlet-multi-silicon-inference-cloud/) and in [inference sizing becoming explicit capacity planning](https://newruntime.com/posts/llm-inference-gpu-capacity-planning/), and it sits next to the control question raised by [Anthropic's model hardware standard](https://newruntime.com/posts/anthropic-model-hardware-standard-control-boundary/).

The numbers are OpenAI's own runs against unnamed comparison systems, and ramp plans are stated as intent. The condition to watch is whether independent InferenceX submissions reproduce the per-watt range, because a first-party benchmark result on a public benchmark is still a first-party result.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Agent Retrieval](https://newruntime.com/topics/agent-retrieval.md): Explore the agent-retrieval topic hub.
- topic hub: [Agents](https://newruntime.com/topics/agents.md): Explore the agents topic hub.
- topic hub: [Ai](https://newruntime.com/topics/ai.md): Explore the ai topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)