@ibab: The team at Ivo is incredibly talented and moves at light speed. It's been a pleasure to help them take their contract …
Summary
River AI promotes its frontier AI development stack, claiming it helped Ivo's contract review agent jump from 70% to 91% on the LAB benchmark at lower cost than competitors, with RL training and inference on optimized open-weight models via River Cloud or on-prem clusters.
View Cached Full Text
Cached at: 10/02/26, 04:44 PM
The team at Ivo is incredibly talented and moves at light speed. It’s been a pleasure to help them take their contract review agent from 70% on LAB to a frontier performance of 91% at much lower cost than competitors.
The tools we offer on http://river.ai help you get up and running with advanced RL and post-training in minutes, and we help to ensure success of your projects. Reach out to us at [email protected] if you we can help you too.
River AI — Your own frontier AI lab. In one API.
Source: https://river.ai/ River / The frontier AI development stack
Build models and agents that learn from experience. Use reinforcement learning to sharpen their reasoning, tool use, and ability to get things done. Train with River, then serve on River Cloud or run the model on your own GPUs.
Trusted by teamsat the frontier.
The River stack
Meet your lab. And your inference stack.
Training
Reinforcement learning for better agents
Training and inference workers form a learning cycleTraining workers send updated weights to inference workers. Inference generates rollouts that are scored, feeding rewards and training data back into training.UpdatedweightsRollouts& rewardsTraining workersInference workers
Inference
Your models, serving your users
Inference serves a large audienceInference workers serve deployed models to many users and applications.Inference workers
River APIOne shared GPU fleet
Training & rolloutsShared GPU fleetInference
Use River Cloud ordeploy on-prem on your own GPUs
Optimized models · RL tested at scaleExplore the system
Training & inference
The next era of AI. Intelligence you own.
Training
Build agents that get better at your work.
Fine-tune an open model on expert examples, then improve it with reinforcement learning. Define the tasks, tools, and rewards; River runs the training and inference that power your learning loop.
Production inference
Put the intelligence you build to use.
Turn a trained checkpoint into a model deployment. Bring your own model’s capabilities into customer-facing products, internal tools, and agents.
Model engineering
Every model, tuned. Every run, tested.
Production inference
Optimization at every step
We optimize every supported open-weight model for training and inference. Our custom implementations run faster than open-source alternatives. We handle GPU topology and performance tuning, and test each model for correctness—so you can compare models and experiment with confidence.
RL reliability
Reliable RL at scale
At scale, subtle errors can undermine a learning loop. We carefully evaluate and stress-test the API to keep KL mismatch and numerical instability under control—so you can focus on improving your agents and developing new training algorithms.
Open-weight models, optimized for training and inferenceQwen3.8 · Dense
Qwen3.8 27B
262k
Qwen3.6 · MoE
Qwen3.6 35B
262k
Qwen3.5 · MoE
Qwen3.5 397B
262k
Kimi · MoE
Kimi K2.6
32k262k
GLM · MoE
GLM 5.2
32k262k
GLM · MoE
GLM 5.3 Flash
262k
DeepSeek · MoE
DeepSeek V4 Flash
262k
Nemotron · MoE
Nemotron 3.5 Lightning 30B
262k
Own the stack
Frontier capability. Sovereign control.
Your models are a strategic asset. Build and run them where you choose—with control over your data, your weights, and your compute.
Managed infrastructure
River Cloud
The complete AI lab, ready to use. Focus on your models while River operates the infrastructure.
- Managed training, sampling, and inference
- Download your trained weights and take them with you
- Usage-based, per-token pricing
On-prem · Your infrastructure
River Cluster
Turn the GPUs you already own into a frontier AI lab. Deploy the same River product in your environment.
- Training and inference on your existing GPU cluster
- Your data and model weights in your environment
- The same API, workflows, and Console
Control the learning
From experiment to deployment.
Open model → your training run
Distributed training from Python
Choose an open model, configure its adapters, and start learning from your data. River handles the compute behind each training step.
train\_steppipelines forward/backward and the optimizer update on the server. Control the learning rate, loss function, and optimizer settings.
train.pyTokenized batch supplied by you
import river_client as river
client = river.Client(api_key="...")
with client.session(project="my-lab") as session:
model = session.create_model(
base_model="Qwen/Qwen3.6-35B-A3B-FP8",
lora=river.LoraConfig(rank=16),
)
model.train_step(
batch, lr=1e-4,
loss_fn="cross_entropy",
)
# Continue sampling or deploy in this session.
The River Console
See what’s happening across your cluster without losing sight of the experiment. The Console brings your team’s training runs, checkpoints, and deployments into one place.
Start building
Everything you need to get started.
Explore the training recipesPython examples · GRPO & CISPOSimplified research examples showing the shape of the API. Bring your own data preparation, reward functions, and evaluation.
**A toy example:**a minimal GRPO loop on GSM8K, training Kimi-K2.6-NVFP4. Reward starts low because Kimi tends to think for longer than the 128-token budget, then climbs above 0.9 reasonably quickly as it learns to answer within the limit.
rl_loop.py
import river_client as river
from datasets import load_dataset
from transformers import AutoTokenizer
MODEL = "nvidia/Kimi-K2.6-NVFP4"
BATCH = 256 # prompts per step
GROUP = 4 # samples per prompt
MAX_TOKENS = 128
LORA_RANK = 16
LR = 4e-5 # learning rate
client = river.Client(api_key="...")
tok = AutoTokenizer.from_pretrained(MODEL, trust_remote_code=True)
gsm8k = load_dataset("openai/gsm8k", "main", split="train") # math word problems
def reward(text, answer):
# 1.0 if the \boxed{...} answer matches the GSM8K ground truth, else 0.0
return float(extract_boxed(text) == extract_gsm8k_answer(answer))
def make_prompt(question):
... # render the question with the chat template (+ "...answer inside \boxed{}" suffix)
with client.session() as session:
model = session.create_model(
base_model=MODEL,
lora=river.LoraConfig(
rank=LORA_RANK,
train_attn=True, # attention projections
train_mlp=True, # MLP / MoE experts
train_unembed=True, # unembedding (lm_head)
),
)
for step in range(len(gsm8k) // BATCH):
rows = gsm8k.select(range(step * BATCH, (step + 1) * BATCH))
prompts = [make_prompt(q) for q in rows["question"]]
# 1. Sample a group of candidate answers per prompt
groups = model.sample(
prompts=prompts, num_samples=GROUP, max_tokens=MAX_TOKENS, stop=["<|im_end|>"],
)
# 2. Score, then center rewards into GRPO advantages (no std division)
train_data = []
for prompt, samples, answer in zip(prompts, groups, rows["answer"]):
rewards = [reward(s.text, answer) for s in samples]
baseline = sum(rewards) / len(rewards)
if all(r == baseline for r in rewards):
continue # no signal — skip this group
ptoks = tok.encode(prompt, add_special_tokens=False)
for s, r in zip(samples, rewards):
adv = r - baseline
full_ids = ptoks + s.tokens
train_data.append({
"input_ids": full_ids,
"attention_mask": [1] * len(full_ids),
"old_logprobs": [0.0] * (len(ptoks) - 1) + s.logprobs + [0.0],
"advantages": [0.0] * (len(ptoks) - 1) + [adv] * len(s.tokens) + [0.0],
})
# 3. One policy-gradient update
if train_data:
model.forward_backward(data=train_data, loss_fn="importance_sampling")
model.optim_step(lr=LR)
model.save_weights("final")
reward/mean · 29 steps
reward
rl_loop.py · Kimi-K2.6-NVFP4 · GSM8K · GRPO · hover to inspect any step
River Cloud Token PricingTraining and inference on River Cloud are billed per token. For deployment on your own GPUs,contact us about on-prem pricing.
Swipe to compare token prices
ModelContextPromptCachedCompletionTrainingQwen3.8-27B-FP8262k1\.80/ 1M0.360/ 1M5\.50/ 1M4.10/ 1MQwen3.5-9B262k0\.66/ 1M0.132/ 1M1\.99/ 1M1.46/ 1MQwen3.6-35B-A3B-FP8262k0\.33/ 1M0.066/ 1M0\.82/ 1M1.00/ 1MQwen3.5-122B-A10B-FP8262k1\.00/ 1M0.200/ 1M3\.00/ 1M4.00/ 1MQwen3.5-397B-A17B-FP8262k3\.32/ 1M0.664/ 1M8\.30/ 1M10.00/ 1MKimi-K2.6-NVFP432k1\.22/ 1M0.244/ 1M3\.06/ 1M3.67/ 1MKimi-K2.6-NVFP4-262k262k4\.28/ 1M0.856/ 1M10\.70/ 1M12.84/ 1MGLM-5.2-NVFP432k1\.46/ 1M0.292/ 1M3\.67/ 1M4.40/ 1MGLM-5.2-NVFP4-262k262k5\.14/ 1M1.028/ 1M12\.84/ 1M15.41/ 1MGLM-5.3-Flash262k1\.50/ 1M0.300/ 1M3\.00/ 1M8.00/ 1MDeepSeek-V4-Flash-0731262k1\.50/ 1M0.300/ 1M3\.00/ 1M8.00/ 1MNemotron-3.5-Lightning-30B-A3B-NVFP4262k0\.30/ 1M0.060/ 1M0\.80/ 1M1.00/ 1M Prices in USD per 1M tokens · Preview rates, subject to change · Cached prompt tokens billed at 20% of the prompt rate Checkpoint storage billed at $0.10 / GB / month · Longer context lengths and additional models are priced on request —talk to our team
Our vision
Intelligence you own, at every layer.
We’re building a future where your AI learns from you, works for you, and belongs to you—from how it learns to the hardware it runs on. River’s training and inference platform is one part of that larger vision.
Talk to our team
Bring your team’s AI to River Cloud or run River Cluster on your own GPUs. Tell us a little about your plans, and we’ll help you get started.
Prefer email?[email protected]
Ivo (@heyivoai): At Inscribe, Ivo’s first user conference, we announced three developments advancing contract intelligence:
→ Ivo Collaborate: Run contracting from intake through negotiation, approval, and signature—with AI reading each agreement, applying legal’s guardrails, routing by risk,
Similar Articles
@AnjneyMidha: many new frontiers will be pioneered @ibab and team are enabling the inevitable
Igor Babuschkin (former DeepMind/OpenAI/xAI) announces River AI, a startup claiming the best system for building internal custom models, backed by a decade of top-tier model training experience and enthusiasm from Anjney Midha.
@ibab: The River API was tested in this blog post and outperformed Tinker on reinforcement learning runs with identical traini…
The River API was tested and outperformed Tinker on reinforcement learning runs with identical training code, featuring improvements in details like routing replay and being open-sourced with ablations.
@ibab: I sat down with Hemant in his garden to discuss River AI's vision and our bets for how to democratize control of the AI…
Igor (ibab) interviews Hemant Taneja about River AI's $1.1B funding round and its vision for open-weight AI models that individuals can train and control.
@ibab: What are the major blockers to build powerful personal AI today and how can we solve them? I gave a talk about River AI…
Igor Babuschkin, CEO of River AI, gave a talk at UC Berkeley about the blockers to building powerful personal AI and shared the company's research roadmap centered on continual learning.
@ibab: We live in interesting times: Superintelligence is around the corner. We believe it is critical to democratize advanced…
River AI emphasizes the critical need to democratize advanced AI training amid the rise of superintelligence and announces job openings to build the future of personal AI.