Singapore just showed up on a SWE-bench chart.
On July 13, 2026, @RoundtableSpace amplified Agnes 2.5 Pro — a free API from Sapiens AI claiming 82.7 on SWE-bench Verified, 78.7 on Multilingual, and SWE Atlas gains that beat GLM 5.2 and DeepSeek V4 Pro on several cuts. The post hit 59.6K views in hours.
The frontier map is no longer US + China only — but read the fine print: these are internal evaluations, the weights are not open, and Opus 4.8 still leads on the hardest bars.
TL;DR — what people are asking
| Question | Answer (July 13) |
|---|---|
| What shipped? | Agnes 2.5 Pro + 2.5 Flash — coding agent models |
| Headline scores? | 82.7 SWE-bench Verified · 78.7 Multilingual |
| Free? | Yes — API at apihub.agnes-ai.com/v1 |
| Open source? | No — API-only, no Hugging Face weights |
| OpenRouter? | Not yet — register on Agnes platform |
| vs Opus 4.8? | Trails on Verified/Pro/Terminal · leads Multilingual |
| Singapore frontier? | Product company, not NAIS sovereign training |
Who is Agnes AI?
| Field | Detail |
|---|---|
| Company | Sapiens AI (parent) · Agnes AI (product/API) |
| Entity | Singapore Sapiens Technology PTE |
| Founded | July 2025 |
| Founder | Bruce Yang — Raffles Institution · NUS AI PhD track · UC Berkeley CS + Applied Math |
| Funding | ~$20M total · $10M Series A (Feb 13, 2026, LOOK FORWARD VCC) |
| Users | 6M+ (Feb 2026, per TechTimes) |
| ARR | Approaching $20M (Mar 2026) |
| IPO rumor | Singapore Exchange listing target end of 2026 |
Agnes positions as world-class multimodal AI for everyone — text, image, video on one OpenAI-compatible gateway. Prior third-party rankings (May 2026): Claw-Eval top 10, Artificial Analysis image top 20, PinchBench top 10 — Agnes claims first Singapore-founded lab at that tier.
Official docs: agnes-ai.com/doc/overview · GitHub catalog: AgnesAI-Labs/AgnesAI-Models (773★ as of July 2026).
The benchmark chart — internal evaluation
Agnes published an INTERNAL EVALUATION comparing Agnes 2.5 Pro, 2.5 Flash, and 2.0 Flash against GLM 5.2, DeepSeek V4 Pro 1.6T, Claude Opus 4.8, and Qwen3.5 397B.
Agnes 2.5 Pro — headline numbers
| Benchmark | Agnes 2.5 Pro | Best competitor on chart | Leader |
|---|---|---|---|
| Terminal-Bench 2.1 | 77.3 | Opus 85.0 | Opus |
| SWE-bench Verified | 82.7 | Opus 87.6 · DeepSeek 80.6 | Opus |
| SWE-bench Pro | 61.8 | Opus 69.2 · GLM 62.1 | Opus |
| SWE-bench Multilingual | 78.7 | DeepSeek 75.2 · GLM 73.3 | Agnes |
| SWE Atlas — QnA | 40.8 | Opus 48.8 · DeepSeek 27.2 | Opus |
| SWE Atlas — RF | 42.4 | Opus 46.7 | Opus |
| SWE Atlas — TW | 38.9 | (Qwen 18.5) | Agnes |
explainx.ai read: Agnes 2.5 Pro is competitive with DeepSeek V4 and GLM 5.2 on several cuts, not yet Opus-class on Verified/Pro/Terminal — but free changes the economics for agent builders testing harnesses.
Agnes 2.5 Flash — generational jump from 2.0
| Benchmark | 2.0 Flash | 2.5 Flash | Delta |
|---|---|---|---|
| Terminal-Bench 2.1 | 52.6 | 62.3 | +9.7 |
| SWE-bench Verified | 72.4 | 75.8 | +3.4 |
| SWE-bench Pro | 49.6 | 50.4 | +0.8 |
| SWE-bench Multilingual | 67.3 | 69.1 | +1.8 |
| SWE Atlas QnA | 15.8 | 36.5 | +20.7 |
| SWE Atlas RF | 11.4 | 29.5 | +18.1 |
| SWE Atlas TW | 13.5 | 27.5 | +14.0 |
SWE Atlas is a newer agentic coding suite — Ornith-1.0 docs describe QnA/RF/TW harnesses with mini SWE agents. Agnes's 2.0 → 2.5 Flash jump on Atlas is the story Mario Nawfal highlighted — if reproducible, it signals real agent-training progress, not just Verified saturation.
Critical caveat — internal eval
The chart is labeled INTERNAL EVALUATION. Unless Agnes publishes:
- Harness configs (OpenHands? Terminus? custom?)
- Temperature, context window, run count
- Independent reproduction on SWE-bench leaderboards
…treat numbers as directional marketing, not field position. Same discipline we apply to vendor GPT-5.6 threads and GLM arena claims.
Free API — how to test today
Quick start (OpenAI SDK)
from openai import OpenAI
client = OpenAI(
api_key="YOUR_AGNES_KEY",
base_url="https://apihub.agnes-ai.com/v1",
)
response = client.chat.completions.create(
model="agnes-2.5-pro",
messages=[
{"role": "user", "content": "Fix this pytest failure: AssertionError on line 42"}
],
)
print(response.choices[0].message.content)
Registration path
- Sign up: platform.agnes-ai.com — email, no credit card for own models
- Create API key in console
- Swap
base_urlin existing OpenAI-compatible agents - Backoff on 429s — free tier 20 RPM text (per GitHub catalog, June 2026)
Model IDs (July 2026)
| Model | Type | Notes |
|---|---|---|
agnes-2.5-pro | Text / coding | New Pro tier |
agnes-2.5-flash | Text / coding | Faster · Atlas gains |
agnes-2.0-flash | Text / VLM | Prior gen · Claw-Eval #9 claim |
agnes-image-2.0-flash | Image gen/edit | Multimodal stack |
agnes-video-v2.0 | Video + audio sync | Async task API |
Not on OpenRouter yet — @skipper4848 asked on X; no public listing confirmed. Third-party models through Agnes require paid membership (pricing TBA).
Agnes 2.5 Pro vs frontier models
vs Claude Opus 4.8
| Dimension | Agnes 2.5 Pro | Opus 4.8 |
|---|---|---|
| SWE-bench Verified | 82.7 (internal) | 87.6 (internal chart) |
| SWE-bench Pro | 61.8 | 69.2 |
| Terminal-Bench 2.1 | 77.3 | 85.0 |
| Access | Free API | Claude subscription / API $$$ |
| Weights | Closed | Closed |
| Multimodal | Native image/video API | Claude vision via API |
When Agnes wins: Budget agent loops, Multilingual SWE, Singapore/APAC latency, zero-cost eval harness prototyping.
When Opus wins: Hardest autonomous coding, Fable-class depth, production where 3–5 pt Verified gap matters.
vs GLM 5.2 (open weights)
| Dimension | Agnes 2.5 Pro | GLM 5.2 |
|---|---|---|
| SWE-bench Multilingual | 78.7 | 73.3 |
| Terminal-Bench 2.1 | 77.3 | 81.0 |
| SWE-bench Pro | 61.8 | 62.1 |
| License | API-only | MIT · self-host |
| Cost | Free tier RPM limits | GPU + hosting |
@preferredev_ on X: "probably built on another open source model like GLM, Deepseek, or Kimi" — unverified. Agnes markets indigenous multimodal training; no public architecture paper yet.
vs DeepSeek V4 Pro
| Dimension | Agnes 2.5 Pro | DeepSeek V4 Pro |
|---|---|---|
| SWE-bench Verified | 82.7 | 80.6 |
| SWE-bench Multilingual | 78.7 | 75.2 |
| Terminal-Bench 2.1 | 77.3 | 64.0 |
| SWE Atlas QnA | 40.8 | 27.2 |
| Access | Free Singapore API | DeepSeek API / weights TBD |
Agnes leads on terminal + Atlas in this chart; DeepSeek still competitive on Verified.
Singapore and the frontier narrative
Our Singapore AI landscape (June 2026) framed the city-state as trusted hub + governance, not indigenous GPT-class training. Agnes complicates that story:
Singapore AI (July 2026)
├── NAIS 2.0 — sector missions, NVIDIA lab, ASEAN chair 2027
├── Hub routing — US/China/open-weight deployment
└── Agnes AI — indigenous product · free API · SWE-bench marketing
Not the same as: US frontier labs (OpenAI, Anthropic) or China sovereign stacks (Zhipu, DeepSeek, Qwen). Agnes is a venture-backed API company — closer to regional product innovation than NAIS compute sovereignty.
Geopolitical angle from X: @UsmanAnzaar — "More countries releasing models = larger attack surface." Free multimodal APIs lower builder friction; they also expand prompt-injection and data-exfil surfaces for agent tools. Same threat model as Grok build repo secrets and DCG for coding agents.
What developers should do now
- Register and run 50 real tickets from your backlog — not leaderboard faith
- Compare against GLM-5.2 self-hosted on cost-at-quality
- Log harness — document model ID, RPM errors, tool-calling behavior
- Do not ship secrets — free API ≠ enterprise DPA; read Agnes security notice
- Watch OpenRouter — routing would change multi-model stacks
- Pair with local guardrails — DCG if agents get shell access
Agent loop prompt (copy-paste)
You are a coding agent using Agnes 2.5 Pro.
Task: {one real GitHub issue from our repo}
Rules:
- Read failing test output first
- Propose minimal patch
- Run tests before claiming done
- Stop if rate-limited (429) and report
Honest limitations
| Gap | Detail |
|---|---|
| Not open source | @BosonJoe: "Not open source, unfortunately" |
| Internal eval only | No public SWE-bench submission link yet |
| Rate limits | 20 RPM free · burst errors documented |
| Architecture opaque | Distillation rumors unconfirmed |
| Enterprise | DPAs, audit logs, residency — verify before prod |
| Multimodal claims | Image/video free — quality vs Flux/MiniMax needs your eval |
What to watch next
| Signal | Why |
|---|---|
| Independent SWE-bench run | Confirms or deflates 82.7 |
| OpenRouter listing | Price discovery + routing |
| Weight release | Would flip open-source narrative |
| Agnes 2.5 Flash default | Cheap agent tier for APAC |
| Singapore IPO | SGX listing end-2026 rumor |
| NAIS update | Does government cite Agnes as sovereign win? |
Related on explainx.ai
- Singapore AI landscape — NAIS hub vs frontier
- GLM-5.2 MIT open weights — Code Arena #2
- Closed vs open-source frontier alternatives
- SWE-bench complete guide
- Ornith-1.0 — SWE Atlas methodology
- Codex vs Claude Code
- China free-model playbook
- DCG — agent command safety
Sources: @RoundtableSpace July 13, 2026 · Agnes AI docs · AgnesAI-Models GitHub · TechTimes — Agnes funding/users · DEV Community — free multimodal API
Benchmark values reflect Agnes internal evaluation as circulated July 13, 2026. Reproduce on your own tasks before changing production defaults — vendor charts are starting points, not warranties.