Pinned Tweet
OPPO × Phala published a joint paper on Kubernetes Pod-level remote attestation for confidential AI.
Goal: prove the hardware, container image, and Pod identity before sensitive data enters the workload.
arxiv.org/abs/2606.03323
Goal: prove the hardware, container image, and Pod identity before sensitive data enters the workload.
arxiv.org/abs/2606.03323
45
39
216
69,797
FLUX.2 klein 4B is live on Phala—served inside a TDX-attested GPU TEE.
Private, verifiable inference for image generation.
phala.com/models/black-f…
Private, verifiable inference for image generation.
phala.com/models/black-f…
11
6
62
3,859
Voxtral Small 24B is live on Phala—served inside a TDX-attested GPU TEE.
Private, verifiable inference for audio transcription.
phala.com/models/mistral…
Private, verifiable inference for audio transcription.
phala.com/models/mistral…
1
1
13
562
Phala Confidential AI: 120.50 billion tokens over the past 24 hours.
Top models:
Kimi K3 69.39%
Xing4.0 29B A4B 10.71%
Qwen3.8 27B 8.61%
phala.com/models/moonsho…
TEE-backed AI infrastructure.
Top models:
Kimi K3 69.39%
Xing4.0 29B A4B 10.71%
Qwen3.8 27B 8.61%
phala.com/models/moonsho…
TEE-backed AI infrastructure.
8
12
77
4,475
Congrats on the launch, @proofableme! 👏 Agents need verifiable identity, ownership, and authority as they act across apps. Proud to be part of what you’re building.
Billions of agents are coming online.
Nobody can tell what's real.
Introducing Proofable.
It lets people and AI agents carry identity, ownership, and authority anywhere.
Proof is the missing layer.
Make it Proofable.
10
14
79
5,269
As inference providers multiply, price matters. Model service quality still comes first.
OpenRouter is raising the bar by rewarding inference that is fast, reliable, and cost-efficient. Salute 🫡
OpenRouter is raising the bar by rewarding inference that is fast, reliable, and cost-efficient. Salute 🫡
We just shipped our biggest improvement to our provider router since the founding of OpenRouter.
Before, we prioritized mostly based on price. The new router now rewards the best quality inference for the actual cost of a request, so requests stay cheap but are also fast and at deliver more value automatically win more traffic, so every provider has a reason to keep improving. And every time a provider gets better, so does your inference on OpenRouter.
4
15
72
4,849
Private AI should be as easy to use as any other API. Together with @foundry_fi, we're bringing Phala's confidential GPU compute to builders and agents through an OpenAI-compatible API, with payments in USDF.
1/ Foundry is partnering with @PhalaNetwork.
Private execution, verifiable results, paid in USDF.
8
16
97
6,166
Phala Confidential AI: 186.42 billion tokens over the past 24 hours.
Top models:
GLM-5.3 Flash 69.41%
deepseek-v4.1-flash-1017 7.61%
xing4.0-29b-a4b 5.50%
phala.com/models/z-ai/gl…
TEE-backed AI infrastructure.
Top models:
GLM-5.3 Flash 69.41%
deepseek-v4.1-flash-1017 7.61%
xing4.0-29b-a4b 5.50%
phala.com/models/z-ai/gl…
TEE-backed AI infrastructure.
19
15
104
7,740
Phala Confidential AI: 186.26 billion tokens over the past 24 hours.
Top models:
GLM-5.3 Flash 59.87%
Kimi K3 13.74%
DeepSeek V4.1 Flash 9.13%
phala.com/models/z-ai/gl…
TEE-backed AI infrastructure.
Top models:
GLM-5.3 Flash 59.87%
Kimi K3 13.74%
DeepSeek V4.1 Flash 9.13%
phala.com/models/z-ai/gl…
TEE-backed AI infrastructure.
15
7
101
6,638
Phala Confidential AI: 105.50 billion tokens over the past 24 hours.
Top models:
GLM-5.3 Flash 43.47%
DeepSeek V4.1 Flash 23.98%
xing4.0-29b-a4b 7.70%
phala.com/models/z-ai/gl…
TEE-backed AI infrastructure.
Top models:
GLM-5.3 Flash 43.47%
DeepSeek V4.1 Flash 23.98%
xing4.0-29b-a4b 7.70%
phala.com/models/z-ai/gl…
TEE-backed AI infrastructure.
17
15
116
9,914
DeepSeek V4.1 Flash 1017 is live on Phala—served inside a TDX-attested GPU TEE.
Private, verifiable inference for long-horizon agents, visual understanding, reasoning.
phala.com/models/deepsee…
Private, verifiable inference for long-horizon agents, visual understanding, reasoning.
phala.com/models/deepsee…
16
15
109
6,058
1M context. Tool use, structured outputs.
$0.75/M input · $1.6/M output.
docs.phala.com/phala-cloud/co…
$0.75/M input · $1.6/M output.
docs.phala.com/phala-cloud/co…
2
3
20
2,151
Phala Confidential AI: 81.33 billion tokens over the past 24 hours.
Top models:
DeepSeek V4.1 Flash 56.30%
xing4.0-29b-a4b 12.22%
qwen3.8-27b 9.39%
phala.com/models/deepsee…
TEE-backed AI infrastructure.
Top models:
DeepSeek V4.1 Flash 56.30%
xing4.0-29b-a4b 12.22%
qwen3.8-27b 9.39%
phala.com/models/deepsee…
TEE-backed AI infrastructure.
17
10
115
8,684
Phala Confidential AI: 57.82 billion tokens over the past 24 hours.
Top models:
DeepSeek V4 Flash 0731 24.69%
GLM-5.3 Flash 24.69%
Kimi K3 14.28%
phala.com/models/deepsee…
TEE-backed AI infrastructure.
Top models:
DeepSeek V4 Flash 0731 24.69%
GLM-5.3 Flash 24.69%
Kimi K3 14.28%
phala.com/models/deepsee…
TEE-backed AI infrastructure.
12
9
93
6,966
Private AI begins on-device.
Case study: how @solrouterai keeps its backend outside the plaintext path and opens encrypted prompts inside an attested Intel TDX enclave on Phala.
phala.com/posts/solroute…
Case study: how @solrouterai keeps its backend outside the plaintext path and opens encrypted prompts inside an attested Intel TDX enclave on Phala.
phala.com/posts/solroute…
14
18
117
7,772
From Claude Code, Cursor, Claude Desktop, or the SDK, sealed requests flow to an Intel TDX enclave on Phala.
The enclave holds the decryption key and protected orchestration. Qwen3.5 122B extends the boundary into an NVIDIA Confidential Computing H200.
The enclave holds the decryption key and protected orchestration. Qwen3.5 122B extends the boundary into an NVIDIA Confidential Computing H200.
2
3
30
2,279
Open-source @solrouterai SDK:
github.com/Router-Labs/ro…
Docs:
docs.solrouter.com
A practical private AI blueprint: client-side encryption, attested workload keys, protected orchestration, and confidential inference.
github.com/Router-Labs/ro…
Docs:
docs.solrouter.com
A practical private AI blueprint: client-side encryption, attested workload keys, protected orchestration, and confidential inference.
0
2
22
614
Agent decisions often touch customer messages, policy documents, and approval state.
Kev-4B is now live in beta on Phala with NVIDIA Confidential Computing.
phala.com/models/jaredpa…
Kev-4B is now live in beta on Phala with NVIDIA Confidential Computing.
phala.com/models/jaredpa…
6
11
104
6,793
Kev-4B model-card benchmarks (self-reported, fp32):
• transfer-v4 locked OOD: 83.8% accuracy (656 questions)
• documents-v1 test: 90.3% (936)
• coverage at ≤5% error: 70.1% on transfer-v4 locked
Methodology: huggingface.co/jaredpalmer/ke…
• transfer-v4 locked OOD: 83.8% accuracy (656 questions)
• documents-v1 test: 90.3% (936)
• coverage at ≤5% error: 70.1% on transfer-v4 locked
Methodology: huggingface.co/jaredpalmer/ke…
1
1
16
758
Kev-4B on Phala:
• Qwen3.5 4B Base
• TypeSafe System One compatible
• 8,192-token validated context
• $0.04/M input; free output
• NVIDIA Confidential Computing
• Beta
Read the launch article:
phala.com/posts/kev-4b-l…
• Qwen3.5 4B Base
• TypeSafe System One compatible
• 8,192-token validated context
• $0.04/M input; free output
• NVIDIA Confidential Computing
• Beta
Read the launch article:
phala.com/posts/kev-4b-l…
0
1
9
517